You've only added two lines – why did that take two days?
mrlacey.com
mrlacey.com
Best part is, I'm almost certain to beat 1500, so I've gotten compliments that it "feels snappier".
A game developer was making a game for PlayStation and they were over their memory limit. They were approaching a deadline but couldn't remove anything else from the game to fit it in memory (or disk, I can't remember). So a senior dev came by, changes the code in 1 minute and everything could fit into memory now.
The thing was that at the start of each project, he had declared a variable of 2mb that does nothing, so when every optimisation has been done and it still doesn't fit, he could just remove that variable and free up some more space.
It was also his way of insurance.
While your trick makes you look good, setting the times to match the budget might be more honest. And when the app slows down you can blame the people who take 250ms to do their part when we agreed to 100ms.
As several other people on HN have pointed out more eloquently, it's the variability that kills you faster than the average throughput.
The 100ms was not about end-user response times, it's referring to internal response times between servers. To make a page in 1 second you can't have 3 different services taking 700ms to respond, even if you can make all three calls in parallel. And if you have to call a bunch sequentially, you need the 75th or even the 95th percentile for those services to be pretty good otherwise your 95th percentile for the entire interaction will be very spiky.
When you do, it can easily confuse folks who aren't deeply involved in your project ("haven't I seen this already?") and can hurt team morale because they never get a "shipped it" moment that feels good.
More to the point: enforcing this rule incentivizes teams to build things in small, shippable components. Nobody wants to be left out of demo day multiple weeks in a row.
They usually expect a fully working demo
I've come to realised this, "true" agile is more like being funny and smart(not that I am either). If you have to tell people you are smart or funny, you probably are not. Ever noticed how smart people(the really clever ones) are just absurdly smart without walking around telling everyone "hey I'm smart", usually the nicer they are the more intelligent they are(yea you get exceptions), same with funny :)
I feel it goes double for "agile-processes" if you have to walk around and tell everyone (management or interviewees how agile your process is, it's probably not)
The places that went "we do all the agile incantations in the book, because that's the only way," well, those actually had a scrum-o-fall culture.
On the other hand if you have a look at the original Agile Manifesto[0] it is a different beast all together. It specifically seems to go against using set processes altogether and basically boils down to nurture organic communication, to adapt and focus on getting shit done.
I suppose the "agile" in the former sense is a compromise to edge closer to the latter, while still maintaining a familiar corporate structure.
EDIT: [0] https://agilemanifesto.org/
If you want to be working with a team trying to be funny, at some point you have to use the word "funny" and "comedy" to make sure you all are actually trying to do the same thing. And make sure the producers and show runners agree that it's a comedy you're making.
Same with agile.
If a country explicitly calls itself "demoractic" in the name, it's likely not.
You must demo the product in a form that leaves the right impression of the current state. If you are painting a picture of a polished product, expect polished expectations. Instead, show the bugs and say "we are still working through this section" Show missing pages, show your work in progress. Show wrong colors. Show potential, wave your hands and tell them to imagine this part working. Don't fake it. If that feels wrong, then a demo isn't right.
If you show it in a form that looks complete and polished, even if you say it is not, how can you expect any other conclusion from the viewer other than "it's practically ready!"?
Isn't this a bit of a fallacy? Not everything can be broken down into chunks of work that fit into a single sprint.
There's a reason I stopped bothering with Scrum a while ago, and this is high on the list.
If it helps motivate a team to divide some four-sprint piece of functionality into two shippable chunks that each take two sprints, I'd call that a win. Customers get something a bit faster even if it's not single sprint-sized.
Many times I can’t tell you what tasks need to done let alone how much time is needed and break it into smaller chunks during planning phase.
If planning has to work you should be familiar with what you are building , with poor information on the bug/ code / stack planning agile is just useless overhead . it works great for yet another CRUD app where you know the requirements to the dot and know exactly how to build or fix not always , most management fail to differentiate
All the reasons in the TFA are also why it is hard to estimate , what and how much time will take .
I mean obviously this only makes sense as a way of organizing a team who are trying to build something exploratory - that’s what scrum is meant for. If you are trying to pursue a solo research project within a team the rest of whom are doing scrum then... that’s not a problem agile can solve.
Identifying and solving something similar to [1] with a team is not simply possible when you plan with agile. I am likely going to go the next item once I mitigate the effect without bothering to dig deeper just because someone is clocking me on a timeline I committed.
It kills all the joy and fun, work becomes boring, this is by design, it is hard to run an organization unpredictably. If only management trusted you to deliver without looking over shoulder constantly ( when the situation warrants it)..
It is not only a engineer's gripe, it applies to management too, the board/ market forces them to be very short sighted, unless you are musk/jobs/buffet it is hard not to buckle to market pressure and invest in longer term opportunities.
The point is not that planning is bad, it can do a world of good in many including unpredictable situations, it is more that blinding pushing a framework especially agile because it worked somewhere and everyone says so, or the manager can't be bothered or won't risk doing something different as the situation warrants.
[1] https://cloud.google.com/blog/products/management-tools/sre-...
Scrum is a collaboration hack for creative problem solving teams, not a managerial accountability tool. The version of scrum where stand ups are for checking on the team’s progress and velocity is reported on up the org is using the tools of scrum to solve a very different problem than the one that it was designed for.
I’m sorry you don’t believe it’s possible but I can tell you from experience that it is possible to use the processes of scrum and the principles of agile to help a team collaborate on open ended creative problem solving tasks.
For all the things that can be atomic like that, it's good practice.
The bigger ones just take time, and aren't shown, until ready for use.
I wasn't talking about intra-team demos to, say, product owners.
Progress gets quantized. Some quanta are small, easily shown, etc... other quanta are a bit bigger, less easily shown.
There is a similar problem in manufacturing.
Atomic releases.
While making something, there are many subtle tweaks to the BOM. Changes, substitutions, removals, adds.
Upstream people can make a real mess out of all that, and one way to prevent it is to only deliver releases that are resolved and intended for manufacture.
"where is revision 4?"
Doesn't exist, won't get manufactured, etc... "Use Revision 5 plz."
For the case of insuring expectations get aligned, a mock up can be used. Deliberately used to generate a spec.
Not having demos of incomplete features would just hide the issues until they are released to the final customers, creating a problem you didn't have before, and making it much more complex to solve.
In my experience everything can be broken down if you spend five minutes actually trying to break it down. And the benefits are very much worthwhile.
People can hide the fact that they have a big ball of mud fir a very long time, and they only want to talk about improvement after hunts have gotten miserable.
4-11 years depending on exactly what you'd define as "brownfield"
> People can hide the fact that they have a big ball of mud fir a very long time, and they only want to talk about improvement after hunts have gotten miserable.
True but beside the point. The same point stands: you can always find a way to make a worthwhile improvement in two weeks - something that's useful on its own, even if it's also the first step of a much bigger improvement plan.
Sure, sounds straightforward enough. Start with simple cases (e.g. universe is a unit circle), you can definitely implement useful pieces within two weeks.
> first the mathematical models behind it need to be created then implemented in software.
That's not a real (i.e. user-facing) requirement.
Sometimes your demo is nothing more than 'here it is in the log doing xyz' or 'I added this thing to this config file'. Not all demos are big flashy ordeals. The team I am on right now most of my demos look exactly like that. I can usually do them right after the standup. Our team allows it because we are mostly remote and talking to each other helps.
I personally use scrum as a weapon to make sure management does not overload our teams. Those made up story points are a good way to say 'you have tasked us with 4 months of work in 2 days'. You have to know your manager too. You have to talk to them. Know what they are looking for. Some take a very hands off approach. Some want the nitty gritty details. For both of those a 'oh that is going to take 3 months' may sometimes work. But it does not give them actionable items to help you. The task broken down into some sort of chunked out work does. Sometimes you do not know. It is OK to admit that. That is when you make a discovery story. Make sure they are onboard with that story is to help you find out what is needed. Even then you will still learn along the way.
I worked with one guy who wanted to task things down to 15 minute increments, 6 months from now. He kept failing. Because he was being too narrow. He refused to do story points. Because they were 'stupid' yet management kept piling more stuff on him to fail at. He was in every weekend and in until 9PM every night. Because he had no tools to push back. Give your management numbers and actionable items or they will assume everything is hunky dory.
It was my first professional development job and I learned a lot of things. One of the keys was this:
Your boss and employer are not your friends. When you promise something to your friends, you owe it without any expectation of reward or return. But your employer owes you something for everything you give them. If they fail to meet their end of the bargain, that's on them and you are absolutely 100% free to leave. That can be money, time off, good working conditions, respect, trust, or any number of things. For me money was not primary (though it is nice): I wanted a good work environment, interesting work, and respect. Fail any of those and I will be looking for an exit, even if it's just a transfer within the same company. But respect is first in that. A boss who shows little respect to his employees is not someone I want to be around. I had that boss who chewed me out, I had another that kept me doing busy work for a year and kept saying worthless platitudes like "you're important to our work, we can't do this without you" when everything pointed to that being false. He was just empire building and trying to grow his direct report count, but it hurt those of us under him because he didn't have the work (but he did have the money) to keep us there. (I was still relatively junior at that job and stayed longer than I should've.)
[0] The hierarchy was something like: VP of Engineering (several divisions below him) -> Chief of Engineer (Aviation) -> Chief of Software (Aviation) -> Chief of Testing (Aviation) -> Me; other production lines had similar chains below the VP.
[1] When she was my manager I quit, or more accurately transferred to another group. Years later, I quit my last job for several reasons, one was her. She was not in our division while I was in that position, but they'd just hired her on. When I heard the name I had to verify, when I confirmed it was the same person I was ready to exit.
Not if you are on an H1B visa ;-)
This works even when you're on a visa, but can take longer to be truly comfortable. I had a classmate in college who'd secured a work visa. His plan was to work for perhaps a decade in the US, save aggressively, and then return to his home country. Salaries in the US were 10x higher at the time than in his home country. Saving even 20%/year meant saving 2 years' of income every year he worked. And he was frugal enough to save a lot more than that. Invested well, taking what he learned back home, he would be set for life at this point (I didn't stay in touch so I only know the plan started well, not how it turned out).
Can you expand on this? I recently moved to the US and people keep telling me to get a 401k, but I haven't yet.
At a minimum, you should save enough to your 401K to get the company matching.
Let's say the company match 50% on the first 6% you save.
So if you put 6% of your salary to 401k, the company will add another 3% in addition to what you did. That's a free 3% raise.
Tell me another way that you can get guaranteed 50% return of your money.
In addition, there is the tax advantage where you do not get taxed on your money until you retire. So you money grow without taxes until you actually need them.
Are there bad 401k? yes, but I think very few. Check the rules of your company. But most of them are very good deals and you should take advantage of them if you can.(Perhaps you could roll over part of the 401k into an IRA first, and then take the rest as 72(t)?)
But one must also 100% expect themselves to evaluate thoroughly T&C, pros and cons before signing up for anything, including a visa.
If he tried to chew me out, I'd just say "That's not how I remember things. I said I was 80% done, and, frankly, it was a mistake for you to pull the trigger on the release without confirming with me first."
I happen to get along really well with the person three levels above me, but I can't imagine dealing with getting direct negative feedback from him. Honestly, more of my conversations with him over the years have probably been about things unrelated to work than work-related; we don't just hop over my direct manager and his boss unless there's a really good reason for it.
And yep, I was actively avoiding that environment too.
I take a different but similar approach when I run into situations where I want to get something small in front of a business user before it’s completely ready: I make sure it’s visibly broken in a way that doesn’t detract from my goal for the meeting.
For example: I have a registration form that I want to talk through. The state drop down will only have 3 entries, one validation error will always display, and on final submit you get a “failed” alert instead of a dummy page.
It lets me walk through the page and get whatever feedback I need, but it feels completely broken so non-technical users expect it to take more time to conplete.
Sucks it has to be this way so often.
The alternative is that they think you are done, adding more pressure, and requiring more wasted time later explaining the mis-aligned expectations.
Does this really imply “humans are broken”? We all have limited resources and I think this could just be a first order prioritization mechanism. Of course one has to be aware of the bias but that’s a different problem.
Yeah it’s all perception, but cultivating the right perception is vital to effective, productive communication.
Polished things have a context of "use", not "evaluate", because we use so many polished things every day, but very rarely have any need to evaluate them (mostly only when we're buying them, or when a repair person asks us to describe what's wrong with them.) Whereas unpolished things are mostly "for" evaluating; it's rare that people use unpolished things (outside of, say, disaster-relief infrastructure.)
it's not a way of gaming the system, but a way of prompting better feedback. For instance, in a Design Thinking process prototypes are more useful when they look rough/unfinished. A more polished piece of work might result in people being afraid to break things.
I use both approaches in my work:
1) demo small, tested bits during show and tells and any meetings where the point is to demonstrate progress. 2) demo large, unfinished and barely stable pieces of work when ideating, trying to figure out the next steps
2) is hard and works only if: - we know what the unstable parts are (because we have tests, so we know the gaps) - we know the audience and how much context they have
"While you mentioned this during the demo, I noted: 1. Your state-drop down has only 3 entries. 2. There is a validation error 3. You get an error at the end.
We need you to fix this ASAP !!!"
Would it be better to show no progress at all until it's completely done? I know agile methodologies tell you to demo regularly, but I'm more and more under the impression that they are to provide progress feedback / reports to management.
Something like https://wiredjs.com/ might work, if you are building a web-ui.
Someone is bound to jot that down and interpret it as a hard commitment regardless of any other details or conditions.
What are they going to do? Fire me? I can go anywhere. They can't find me anywhere.
I've learned this adage years ago (I think it was from Spolsky), but nowadays I'm in a project (re)building an UI from scratch as a sole developer aaaand I made the same mistake.
I was doing some UI prototypes about activating a process, big green Activate button, opens up a confirmation dialog, spinners with some artificial / simulated delay because I didn't do the back-end, and it caused confusion with our tester because she was wondering if activation actually works.
I've got three options; remove the button for now (I should do that), partially implement the activation (changing a status in the back-end), or fully implement the activation (which has a lot more prerequisites).
(Cause another lesson is that people have a lot of trouble giving worthwhile feedback on a verbal/written description of something, they gotta see a thing in front of them).
I've learned the opposite. If I communicate clearly that what is being shown is little more than a mock-up, executives of all levels and technical skill, all the way down to managers just above myself, all understand that a very thin and scripted demo is not anything close to a finished product.
Describing the demo as a "house of cards" that will collapse with a single misstep gets the point across nicely, while also giving the demo audience an eyeful of what can be accomplished if everything is handled appropriately.
Demos are carefully scripted and rehearsed, values are hard-coded, and absolutely nothing exists that does not prop up the demo for the purposes of the script and the talking points.
It is hard to describe to someone who hasn't written an application demo like this just how little actually exists behind the UI.
Anyway, my point is that if you choose the correct words, anyone can understand that it's like a painting of an application, and not an actual application, just like a painting of your mother is not actually your mother.
Even people who ought to know better don't understand this.
> “I don’t know what’s wrong with my development team,” the CEO thinks to himself. “Things were going so well when we started this project. For the first couple of weeks, the team cranked like crazy and got a great prototype working. But since then, things seem to have slowed to a crawl. They’re just not working hard any more.” He chooses a Callaway Titanium Driver and sends the caddy to fetch an ice-cold lemonade. “Maybe if I fire a couple of laggards that’ll light a fire under them!”
0.https://www.joelonsoftware.com/2002/02/13/the-iceberg-secret...
Management tip: Make the UI reflect the actual state of the project.
The UI should be UGGGLY and should get prettier as the backend work gets finished. If if the artists prettify it; make the animations and interactions janky and sluggish.
Never make the UI better than the actual implementation.
Bonus tip: Always have something slightly off in the UI that management can point out to fix. Useful managers will simply quickly point it out and move on to more important problems; useless managers will focus on it.
For instance, Bootsketch: http://yago.github.io/Bootsketch/
I wish some of the bigger CSS frameworks would adopt sketch theming support as a "progressive" option, rather than the entire page all or nothing being Bootstrap or Bootsketch have the ability to add a "sketch" class to any element on any page. (Or maybe better yet, forcing sketch styles by default and needing something like a "final" or "final final" class everywhere, like the documents folder of someone that has never understood source control.)
In software we can put a few shapes on a screen in just the right context and people will believe there's an operation as sophisticated as the wite-out bottle factory behind it.
Maybe as part of the demo, you should demo that it *conspicuously fails to actually do anything", to help reset unrealistic expectations.
Unfortunately communication for software projects isn't often discussed or considered valuable in this community but you can learn it like anything else in tech.
The hardest language to learn is the one that communicates with people not computers.
Well, after a while they understood, there is a small difference, between a website - and a virtual market place.
Seriously, it is easy to forget, that for most people, all these technical things - is just dark magic in a black box. Which sometimes work and sometimes won't. And I find this sometimes hard to deal with, because society gets more and more technologized. At least a very basic understanding would be helpful.
I also point out when a requested feature actually exists as a whole company.
Just make sure you have infinite money first.
Pointing out that Product XYZ has 3000 employees and has been carving out a niche since 1995, while there are six of us with no knowledge of that market, is usually the only thing that gets him to accept that just because he understands what something does, it doesn't mean he understands what it takes to build it.
This is how we used to sell drugs in college. It was a simple URL you could go to with pricing and you just sent texts to a burner phone to arrange a transaction.
But a virtual marketplace .. where different actors make transactions, is a different story. Consider you have a bug and people loose money because of you. It really needs to be solid.
Maybe they are just looking for a side hustle.
We came to this rule after far too many incidents where some sort of mock up (Photoshop, HTML, whatever) was shown and taken as done. Then we got the questions (possibly unhappily) about where it was and when it would be done because obviously ”it must be” since we “showed it”.
The rule served us incredibly well for years. It put a hard stop to all miscommunication. Everyone understood exactly where we were in the project. Either we knew what it was going to look like, or (if it was ready) it was done and awaiting their approval.
Shortly before I left we got a new graphic designer. He wasn’t embedded with the programmers. He didn’t know THE RULE. Sure enough, we were asked why a design we’d never seen before wasn’t ready yet. Because he made a mock up in Photoshop. He told them it was a mock up. Doesn’t matter.
We basically follow THE RULE at my current job. It still works wonders.
We sometimes have to do that in technical reports. The proofreaders always feel like they have to give at least one comment, so you give them a clear mistake to point out to avoid useless debate over minor points.
Whether we realize that immediately or only after we stopped having misunderstandings I’m not sure.
But once it was realized I don’t think it took very long at all for it to become a rule. It made life so much easier for us developers.
As he simply was the marketing/sales guy, in case he found himself cornered he could always pretend that the failure was on the dev/technical side... "see with the support".
He could also successfully sell himself this way: once a billionaire proposed him to be the sales director of some company. He set the meeting at 7AM at the Ritz bar, and got the job (not for long : he had negotiated to keep the perks when leaving anyway).
Superimpose the word "MOCK-UP" in a gigantic font that takes up nearly entirely the image.
Make it translucent and red or make it black outline. And maybe tilt it diagonally to catch attention and to make it easier to visually separate it from the rest of the image.
To streamline the process, you can just keep an image like this around to import as a topmost layer into other images.
So I very quickly used MS Paint to mock it up... Just so I could clarify that is what they wanted.I shared my screen and someone said "great, you've done it!". Even though I was clearly showing a screen that showed me editing a screenshot of the UI, in MS paint...
MS Paint.
sigh, they don't tell you in University that the biggest skill you will need in this job is patience and learning how to channel your inner zen.
Developers don't need 3 monitors to get through the working day, they need regular sessions with a psychiatrist.
Amen. I can see this as dev perk in job ads.
I have yet to see secure software... does that even exist?
You say this in present tense which makes it seem like a generalization. This is also a mistake that non-technical managers (and others) often make. The frontend frequently rivals and sometimes exceeds the backend in complexity. It is very application specific.
No chance for suckass product guys to push if they can't understand what's going on.
I think we spend too much time focused on the trigger man. Whatever person in your org is making your life difficult, there’s a person above them who knows and hasn’t done a goddamn thing about it. Who is the real problem?
I dig out my "The Dilbert Principle" book and will start reading it again to promote sanity.
It's not fair for managers to assume, but it's definitely fair for them to ask. And it's our responsibility to show and explain.
There's the famous story of the first iteration of Gmail being done really, really quickly. The demo was the product. And then just iterations from there. Definitely a good model if possible.
We've all worked with the perfect is the enemy of good guy who deeply considers all aspects, takes 10 times as long to deliver and then eventually, after much blood sweat and tears, delivers equally as bad software as the rest of us.
Personally, I've gotten over myself and try to just ship it.
That said, this is a bit different from the gnarly bug type scenario of the OP; though I'd probably ask how they wrote tests in 2 lines of code? :)
The truth is that in a lot of cases nobody has ever bothered to explain it to them. What I tend to find in a lot of situations is that managers often prefer younger and/or less experienced developers because they can be bullied - but the reality is that this also means that people are then unlikely to tell their managers what they need to know. Ultimately it is the managers creating the problem, but in most cases (mind you not all) the managers don't understand that they are creating a problem.
Not screens that look like doodles, but actual paper and cardboard and maybe some Blue Tac or paint.
Ordinary users apparently behave very differently, because it's obvious that a piece of paper can be changed and they know how to do it so it only takes a little nudging to find out what the customer actually thinks the system should look like.
She still has all the CS background to estimate that a change maybe that seems simple to the user just isn't viable, but using the paper prototypes encourages users to leave that to her and not second guess themselves into accepting a bad design because they're mistakenly assuming it would be hard to make a change when actually this is the perfect time to make such a change.
Now my boss is happy that I’ve demoed API integration, but every next meeting he asks me why nothing has changed. It’s because I’m still waiting for something that is done.
For some reason the fact that it’s not done is always the fault of the front end team. I’d love to get an actually finished API for once...
But, as you describe, from a perceptions stand point, it was the worst thing I could have possibly done. It went from a very happy client, to a very unhappy and confused client when progress "stopped." I actually started recording development work as a way to make them understand all the invisible stuff which goes on behind the scenes.
"Why/how is this worth X dollars/time? I know someone who says they can do it in a week." To which, I eventually learned to reply: "Wow, well... In that case, let me shoot you an article on how to build a Twitter clone in 15 minutes. [awkward pause while I smile at them] There's a lot more than just literal lines of code that goes into building a successful software product."
It was an old Audi S3, like, 2011 model. Had a guy tell me the car was in better condition and the asking price was less than what I had listed. My car was listed at $10K AUD and that car was listed at $8K.
Them: "Why is your car listed at your price, and not matching this car?"
Me: "Well for starters, that cars in Adelaide, we're in Brisbane. If you want to go to Adelaide to check that car out and find out it's been in a bender and had most of it's body fixed and the listing avoids that. Be my guest."
Them: "I doubt that, I think you're asking too much. Will you match that price?"
Me: "No."
Them: "Why are you wasting my time, I should buy that one just to annoy you.'
Me: "You fucking do that then, laters"
I sold my car the week after this for the price I wanted. Straight up https://www.reddit.com/r/choosingbeggars material.
LOC is a decent measure, but features are our targets
Only if your measure of success is tied to fewer LOC.
Ones target should be shipping features. If you use 10k LOC to get a feature out, or 500 lines of more concise, optimised code, what matters is the feature.
If you have LOC targets to meet you are incentivised to produce the former rather than the latter.
My point is that a high number of lines isn't as important as good features. Though the two can get conflated
How often do you actually receive quality bug reports at work? My experience is that external or internal users almost never provide sufficient information and you as a coder are always expected to drill down on what they reported with a barrage of questions.
ie. if you are not doing https://en.wikipedia.org/wiki/Five_whys then you might be doing it wrong and wasting time because of it.
I'm referring to this:
> Some developers would have immediately gone back to the person reporting the problem and required more information before investigating. I try and do as much as I can with the information provided.
Which seems like being stubborn and making a mistake because of it.
Couple other parts also seem a bit overdoing it:
> Because I investigated if there were other ways of getting to the same problem, not just the reported reproduction steps. > Because I took the time to verify if there were other parts of the code that might be affected in similar ways.
These seem like taking a gamble. Maybe something comes up, but is it more probable that this work should be minimised until there is more proof of "other ways of getting to the same problem"? Developer time is expensive, is this really the best way of using it? Would it make sense to just fix the issue at hand and only put in more time if more bug reports come in after the fix or if there is some other indication that this part of the code might be more broken?
I know I'm guilty of this one, and I've stayed away from high paced jobs and appreciate jobs where people are ok with my reluctance to bother a lot of people even if that means it takes me longer to figure things out on my own.
This also means I build a much deeper understaning of the systems I work with, or at least I like to think so, and some people have confirmed that about me, indirectly, by praises of my insights.
Of course I know nothing about you so I'm not trying to give advice, just replying about the "confirmed" being a potential cognitive bias.
> is it more probable that this work should be minimised
I guess this is where automated tests can come in. You fix something and see if passes the unit tests. But then everyone has their own approaches. For him, fixing a similar bug twice is worse than finding all possible mistakes at once.
Maybe you were expecting the dev to go through all the passwords and fix them so the first letter of all passwords is lower case? Oh but if they were following best practices they don't know the password, they only know its salted hash.
Since the app was already shipping without the first letter being auto-lowercased that would suggest there were plenty of passwords with the first letter already upper cased, also something you can't test for easily if all you have is salted hashes.
PS. Yes, we were hashing the passwords.
* smoke means fire
* a contained smokey fire is sufficient to hide the start of a wildfire
This means:
* keep your errors at 0. If it "can't be kept at 0" you're either too far gone or thinking about the issue incorrectly.
* user complaints are errors. Just because they aren't clear doesn't make them any less so.
There is a perception that users go out of their way to make unfounded complaints. In my experience, getting any complaints is the issue.
There is also a perception that some errors aren't important. If you have a channel to recieve an error its because it has business value. If a dev I was managing ignored a p/w entry bug due to non-dupe / assumed user error without significant digging and user interaction I'd be livid. Most businesses will lose significant value if users perceive the act of logging in as difficult.
Getting complaints from users is a good thing, but yeah missing them or not fixing them when you come across them is. As a user, I would be overjoyed if I reported a small issue, and the company fixed it quickly.
Being livid is natural, but I was pretty sure the developer himself knew he screwed up this time, so no point expressing it. Plus as I explained in the earlier comment, this escaped for so long because we did not have many cross device users. So did not affect that many users. Just for my experience, I check this every single time i test a website before it goes live. (and also check with other keyboards than just Gboard)
Okay, how do you handle a network failure, a full disk or a faulty RAM stick?
I see your overall argument, but at some point you've got to accept that you can't handle everything.
So if you're small, a RAM issue is something you deal with manually and rarely. As you get bigger you'll transition to automated failovers that still get looked at individually. Then you'll scale up to the point where these aren't freak occurrences. Now its important for you to have a strategy to identify the issue and its follow on issues, and resolve them. Its also past time you think deeply enough about your setup to be able to contain them so you can stop surfacing them as errors - they are now part of a normal business process. You want to make sure that too many are surfaced as an error (and too few), and any effects you can't currently recover from automatically are also errors.
Its perfectly possible to "keep errors at 0" without ignoring any output.
I don't work at Microsoft.
Very, very often. I work as a QA engineer whose main responsibilities is to go through the bugfixing queue and add needed info where necessary. And I have to spend a lot of time every day doing this. Sometimes it gets so bad I have to assign the ticket back to the reporter to add more info, because even I don't know where to look without it.
Interestingly enough, it's always the more senior people at our company who are guilty of writing crap bug reports.
Half of my current tickets barely have 2 sentences in so-so English.
Edit: And they quit right afterwards.
Telling someone that it only took me an hour to eliminate an hour of someone’s work every week doesn’t go far.
Even worse, someone else now has to explain "why they did it wrong to begin with" regardless if the replacement technology existed or not at the time of original process creation.
Also making more money is better, because it means growing the company with more employees and more job safety for everyone.
Most companies aren't like that though. Know your company!
Also consider if you can run out of money before growth kicks in...
The maths is simple. If my margin is 1%, and you make me an extra $100, I keep a buck. If you save me $100, I keep that whole $100. If I'm smart, I can use that to drive my prices down, and take more market share without reducing my margin.
Obviously it becomes less clear if your margin is >20% or so.
From what I understand a lot of product teams at FB have nearly frictionless development tooling for their use cases so the pressure is to produce volume.
[1] Unless there are multiple skeezy companies in Menlo Park.
Now if there's 5 drawings with different colors, or asking "what color do you want it?" leads to a 5-week email chain...
I don't think you grok what's happening. Some middle manager sees a shade of red on his screen in a PDF, and the dev is expected to reproduce the content in that shade of red. There are simply too many variables.
Even if you have access to the PDF, the red will often be rendered differently by the browser than it is by the PDF engine.
I've had middle managers tell me to "fix" a web site because the colors looked different on his office CRT than it did on the laptop screen of a person in another building.
trivial to fix
Everything is trivial when someone else has to fix it.
I discovered that bug report after our designer noticed at a glance walking behind me (I use Firefox) that the colors on our site were far darker than she intended, somewhere around 2017-2018 (that bug was opened in 2010).
I've left companies because of devs like that. People who just stand in the way of getting the software to do the correct thing. I do not understand what makes these people tick.
No, No you don't.
I'm curious if this is "everybody knows this person isn't doing any work, here's a blindingly obvious metric to use to defend this move to HR" because HR orgs often hate things that don't have metrics attached, or if it's "I don't understand what this person is doing, and I don't see any commits, so it must be nothing"?
They were a senior engineer that spent a lot of time coaching the rest of the team, and less time on their own personal work. The commits data point, in my opinion, should have been the beginning of a conversation that ended with the manager asking this person to adjust their priorities. Instead it was presented as an accusation that the person wasn't getting their work done, which ultimately lead to them feeling insulted and quitting.
That could be completely wrong, of course. But would it surprise you?
I've been fortunate recently to not work in orgs where mentoring and collaborating like that is looked down upon - instead, it's encouraged - but my ongoing struggle is to figure out how to quantify it.
I've got two similar but distinct motivations for wanting to quantify it:
* If I have a manager who looks less favorably on it, I want to be able to demonstrate my worth; but also,
* If I'm spending most of my time trying to mentor the rest of the team, I want to see how well I'm doing - and I want to be able to change things and see if it results in positive or negative changes
Sadly I've completely failed at coming up with how to quantify this so far. It all comes down to qualitative peer feedback/manager feedback...
[0] As a technical person, I try to practice the reverse of this: if someone asks for a feature or claims there's a bug, I try to dig until I fully understand why they're asking for what they're asking. So I think asking for similar curiosity and depth from a non-technical person is fair.
Assuming there's some kind of constant ongoing engagement with one person or group, I'd expect an immediate dip in velocity as your productivity goes elsewhere, then it growing, and maybe evening out to before-engagement numbers (approximately), as they reap the benefits of learning from a senior engineer. Then, as you're able to return to normal duties and they're able to apply the lessons, you should see velocity greater than before the engagement. That delta should be somewhat quantifiable and, ignoring other variables, should represent the benefits of coaching.
There's also huge value in the increased job satisfaction for both mentor who enjoys mentoring, and a mentee who is learning. That should show up in any kind of employee satisfaction survey, or retention numbers.
If anybody can push commits to the repo, then it's a useful metric. Finally, this sort of action should be take after the manager has worked with their report, by having somebody else help them, or put them on a different project.
The smart thing to do is to regularly keep your manager updated on what you're doing, especially if they don't come by regularly and ask you.
Especially if you are WFH.
Meta-metrics might be much more helpful, though harder to come up with, quantify and monitor. Things like defect rates, user reported incidents, user satisfaction, stuff like that.
Things that cannot easily be gamed from within the development process and that are still directly linked to the success and economic viability of the product and its development methodology (though on a higher level).
Code is often improved by removing code.
Task completion would also be abused, because it's too vague. It simply shifts the burden to the one formulating the task (e.g. preventing holes in the specification like missing performance or hardware requirements).
Exaggerated example: if the "task" is to automatically deliver a report containing certain data and formatted in a specified way, the easy way might be an implementation that stalls the DB server for hours with deeply nested FULL OUTER JOINs on non-index fields.
The task would be completed quickly and arguably correctly, since neither runtime nor memory requirements were explicitly specified...
But you said it yourself, any system is going to be abused...
Out of sight, out of mine <== don't let that happen to you
I don't believe it is any coincidence that the highest compensated engineers I know also are highly visible through their own efforts.
The problem is lack of trust in you as the expert and possibly a lack of self-awareness (they think they understand).
A manager/client should make mostly strategical decisions like: We should solve this problem, here are the resources. And almost never tactical.
They also shouldn't even care or look at LoC. They shouldn't be worried about metrics of 'effort' at all.
I found your phrase "lack of trust in you as the expert" a little jarring because (again, in my personal experience) considering the developer to be a domain expert is somewhat of a foreign concept. I suspect/hope the situation is better elsewhere in the industry. :-)
> I suspect/hope the situation is better elsewhere in the industry. :-)
I do freelance, client work. Alone and in (very?) small teams. Typically my/our clients only have a superficial understanding of everything technical (if at all). Trust often needs to be earned.
One way to gain trust is being creative/optimistic and explaining feasible possibilities and strategies. Another one is being pragmatic and not selling them something they don't need, or might not need.
And then the most important one is to have conversations about their problems and wishes. Showing that you understand them by asking questions and writing a specification. And explaining your (iterative) workflow: "Let's figure this part out after we've done this other part." I guess this is the "domain" part of the process.
My experience is that if trust is in danger then the work is less valuable, less fun and less sustainable. Indications of this are things like we discussed before and similar:
- Trying to measure effort instead of rewarding value.
- Nitpicking, bikeshedding and other distractions.
- Overstepping their expertise (typical for UI design, a bit less for programming)
Now most of my interactions are good but sometimes I get the above. We're actually discussing of doing more upfront communication work in the offers and initial discussions to prevent these things (even by filtering out clients/collaborators) and to set a tone. Because again, this is unsustainable on multiple levels and it never ends well...
My leader isn't a technical person by a long shot. Instead they focus on getting people that they can TRUST on their team. Yes, sometimes it means we have to go back and 'fish' for metrics to throw the business. But we do notice that the less we chase metrics (and, yes, the arbitrary goals set out by the company as a whole) the more productive we really are.
In 2014 I worked at a company that switched to Git and then started measuring LoC to assess performance/involvement. Engineers took to committing/removing things like node_modules directories to make the data meaningless.
It still happens, even today, quite a bit.
jaw dropped. since then i'm committed to being the biggest office space corporate schmo possible. let me cog it up, keep payin me 6 figs, lettin me work at home full time.
I agree with you 100%.
Spot on. Talented engineers can usually be picky and change job if they don't like the environment, but not everyone has this option. There's plenty of developers who are stuck in shitty companies (lack of skills/experience, or struggle with interviews, or just live in places with limited opportunities). And the longer you stay in a bad place, the harder it gets to escape it. 2 years ago I was the hiring manager for a few open positions and I was honestly shocked by some candidates. So many "senior devs" that despite having 5-10 years in the industry wouldn't have passed the interview even if they applied for a junior role.
It's very easy to have a distorted view of the industry if you are privileged enough to have only worked in great tech companies. I'm guilty of this myself, making friends with other devs in my city was definitely eye-opening for me.
The business I worked at was a typical office, like the one you saw in Office Space. Departments had their own TV screens on the wall that showed performance of individuals in a department; the sales department had a screen that showed who was making the most sales that day.
After we'd pretty much finished working on the web apps that supported these TV screens, the CEO met with me and a colleague to tell us how good a job we did. Then he said:
"You know, you're the only department now that doesn't have a performance monitor. Maybe we out to get one for you. We could base it on lines of code."
My coworker and I were speechless at first, but we started laughing because we thought it was obvious that he was joking.
"What's funny? Why are you laughing?"
We quickly stopped laughing when we simultaneously realized that our boss was not kidding! I said we'd get right on it in the next sprint, and he told us that sounded good and left. I'll never forget that look on my coworker's face.
Poisoning the well, I like it.
Whats nice about the proceduralism is you can document that the steering committee only meets once a week on Tuesday afternoons and change control meets on Thursday. And everyone knows the automated test suite on DEV takes about half a working day. So if a change can't be worked into the schedule in less than a day, it'll never pass CI testing before the change control meeting so it'll take more than a week.
Whats bad, is mgmt would like you to complete multiple changes perhaps at the same time which always complicates the change control process especially if change #7 failed last week so company policy is to roll everything back and now we have 13 changes, two weeks worth, to complete next weekend. Also whats bad, is knowing its a corporate nightmare to make any change, why did I make a mistake to begin with of having the buttons swapped or a misssspelling or whatever.
I find the big metric now a days is backlog. Lets see the number of request tickets decrease this week instead of increase. That leads to intense pressure to roll multiple problems into one ticket.
This made me realize that working in a purely engineering team can sometimes be a perk. Not because technical people are "better", but because it leads to fewer frustrations like this one.
There was a time when I thought this video was funny:
These days I can only look at it and think "the expert is terrible at calm confrontation and good communication. There would be no problem if he had developed those skills."
It is important to rule out miscommunications. Contradictions often are not obvious to the requestor. It also helps to understand, which of the contradicting requirements can be dropped to resolve the problem. Sometimes the problem is just a small feature which made it to the requirements, because no one considered it a problem.
That doesn't mean, I haven't gotten requests which were as silly as in the sketch and had discussions along the lines shown :) And of course, I never hesitate to speak up about actual issues.
Getting sensible requirements is not only on the expert.
But the expert is the one who can know if the requirements are sensible.
In requirements gathering, the whole job is to hear people's attempts to describe their problem and figure out what problem they actually have.
By definition they don't have your expertise, or they wouldn't need to talk to you.
So, of course they will say contradictory things and use terms completely incorrectly - they cannot do anything else. That's why they hired you.
The expert in this scenario gets hung up on their incorrect language and gets flustered and stymied, telling them "What you want is impossible!"
What they _said_ is impossible. It's our job to persistently, patiently, calmly help them understand their needs, without judging them for needing our help.
I'm not particularly good at it, but I understand the mission.
They absolutely use lines of code metric at my company. I don't miss any chance to tell my manager it's complete bullshit. His answer: "Engineers are supposed to write code, just like construction workers are supposed to build houses."
If you give a construction worker a design to build a wall, and the worker is given 2,000 bricks that must be used to build the wall, then, yes, of course the worker must lay down all 2,000 bricks to build that wall. However, if I am asked to build a computer-simulated model of said wall, and if there is a way to build the model with 200 lines of code that looks and performs identically to a model built with 2,000 lines of code, then, yes, of course I am going to build the wall with 200 lines of code.
I hope you find better pastures.
And I can assure you, they were not judged on the number of views or pages on their drawing, or on the number of variables in their structural calculations.
It took 23 minutes for that rule to dissappear after my lines of code metric jumped 250,000k
$JOB-2, admittedly about 4 years ago now, the good manager with a background in software left for a better opportunity and was replaced by someone who's background was management. With no insight into the subject matter, they fell back on whatever they thought they could quantize.
Got numerous things like that, though my team lead and our project manager did a great job of shielding the team from that crap, we still occasionally hear it come up in group meetings and the like.
I even got a task handed directly to me, bypassing everyone above me, to "estimate how much it would cost to migrate all those linus apps your team has to windows. They'll run better there". Just the windows licenses alone would have cost us about half the existing server costs since we were using AWS instances. Also included a line item that included recruitment costs for a new developer, and verbally informed him that it would likely involve hiring a new team, as the existing team was hired specifically as linux developers.
I was on a browser team. A fellow co-worker decided to add the Fullscreen API to it which means not just add the API but first discuss it in the relevant standards committees.
I'm pretty sure he thought, and so did management, this would be a 2-3 month project at most. IIRC though it was like 18 months, maybe longer.
Some problems that aren't obvious at first
* What is fullscreen mode? Is it a mode for the page, a mode for an individual element? What?
They eventually decided it was for an individual element
* What happens to the CSS above that element when none of its parents are being rendered?
I'm sure that took a while to argue over. Like if it was position: relative or absolute and suddenly it's parent is no longer displayed. What if the parent has CSS transforms? Okay, you say we ignore those and consider it the root element. Okay so does that mean none of the other styles from the parents apply like color or font-family? If some do and some don't we now have to go through every CSS property and decide if it does or does not continue to inherit. I don't actually know the answer to this.
* You have a DOM A->B->C->D->E. C asked to go fullscreen. While there E asks to go fullscreen. User presses ESC or whatever to exit fullscreen. Should it pop back to C or A? Does it matter if E is a video element and they clicked fullscreen? What if C is an iframe does it matter?
* Testing across all devices that support it requires all new testing infrastructure because going fullscreen is effectively something that happens outside the page not inside so testing that it actually happened, that a user can actually exit it correctly, requires entirely new test systems that were not there for previous APIs. Then multiply by 5 at least (Windows, MacOS, Linux, Android, ChromeOS, ...)
And so even though I'm sure everyone ended up understanding that, it turned out to be way more work than anyone expected. Yet, in the back of their minds it was arguably always "this is taking way longer than it should, goals not met" or at least that's how it seemed to be perceived.
~ Edsger Dijkstra
PHB: You've only added two lines - why did that take two days!
Dev: Because it took that long to understand the requirements, understand the code, find the root cause, write a test to verify the root cause, and then 2 minutes to write the code to fix it. The resulting code verifiably fixes the root problem for at least reasonable inputs, with minimal code and thus minimal chance of a bug.
PHB: Why not add 50 lines of code and be done in an hour?
Dev: Because that would be enormously more likely to be buggy.
PHB: But it would fix the symptom, right? Don't let perfect be the enemy of the good!
Dev: 50 lines would not to be "good", it would be crap. And we would have to revisit that code in a week or a month.
PHB: But in the meantime we will have delivered, dammit! We're on a deadline, with three managers and five customer representatives breathing down my neck, and you're telling me you want to spend two days on something you admit could be fixed in an hour?
Dev: I never said I could, only that 50 lines would be a crap solution.
PHB: Are you calling me a liar? Or just an idiot?
[Continue until the developer is fired or gives in, loses hope in humanity, and starts planning their exit.]
Thankfully I have never worked with someone like that, but if the starting point is as bad as the first sentence above then it's only going to go downhill from there.
Exact quote from my manger, several times. They just don’t get it. And then a trivial feature takes a month to more or less implement because it’s all built on a toppling tower of crap.
If situation is that far degraded, the answer is "IDK, does the distinction matter?" And it doesn't: either the PHB is putting words in your mouth on purpose, or couldn't hold 30 seconds of context.
We're going 'fairly functional' versus 'broken'.
Please tell me how to find a place like this?
I think the company is going through a slow-motion death spiral as they try to combat this ("let's do design! DOCUMENT! That's the key!") but it just results in deadlock in the stages prior to coding since the "designers" don't know how any of it works, so push garbage onto the coders to "implement", which just gets pushed back as it's nonsense, and nothing gets written. And permanent complaints over how long everything is taking to write and document. So just bullying of the developers with shouting and swearing - that'll get things written, right??
You were also expected to estimate how long things would take to write (eg expand an existing structure) without an understanding of how anything worked, and then held to it. I was literally asked by a "manager" once: "Why do you need to know how it works? Can't you just write it if you've been told what to do?". That sealed the "let's get the hell out of here" at that point.
Meanwhile "management" couldn't work out what was "wrong with the developers". "And why are they all leaving? Why do they not know what these millions of lines of code do without any explanation or time to look at it? They must be RUBBISH developers. We need better developers. Employ more GOOD DEVELOPERS. What? Isn't any of this documented????! It needs documenting right away! Do documentation! Who knows how to document this? Ahhh the developers. Developers - DOCUMENT THINGS.", then "Why is there no code being written? What? Why are you documenting??? It's taking too long! Write code! And document! At the same time even though it's impossible! WE NEED MORE DEVELOPERS! Where are the GOOD DEVELOPERS???! We need to write a new policy that the developers should adhere to."
Always the developers' fault apparently. Absolute stupidity.
This example is just one case/symptom of much larger problem.
Non-tech managers will only see fast progress on combination of: - poor developer - doing fast progress - with shitty code - on good quality codebase
They will see tech lead/good coder as asshole in general, with poor performance in general, who for some unknown reason is respected for their code and sometimes magically ticks off hard problems quickly which "must be a coincidence, overestimated work to begin with or something like that" on combination of: - person who actually cares about the project - who repairs shitty code/tech debt - who thinks more deeply about the problem - and as a bigger picture issue, not just ticking off ticket with the lowest resistance possible - writes good quality code - if the problem breaks current abstrations, refactors abstration itself - who cares about readers of the code
In a lot of ways what you are describing is how it should work, that the tech lead works on the hard problems and big-picture problems, like abstractions and architectural issues, as well as mentoring junior devs.
In my opinion any manager (especially a non-technical one) should only measure the team as a unit. This is particularly important when evaluating the performance of a tech lead.
One thing I like to look for is natural lines of conflict in a situation that can arise when different individuals are working towards different goals, and question the underlying reasons why a manager may be acting the way they are. In a lot of cases you can get to a win-win situation if everyone is willing to play ball. Of course if the conflict arises from a fundamental organisational flaw, like poor management methods, or poor company culture, then it is time to move on.
1. "Can do" attitude 2. Never backs down from a challenge 3. Being the glue that holds the company together 4. ...
As you can see these are all the kind of things you can't put your finger on but you "know it when you see it". When they see it, it nearly always looks like their reflection.
It always gives me a pause, and then I would double the schedule.
In other words, if a project manager sets an expectation that a task is a “task that takes two weeks”, the developer will somehow manage to use about two weeks to make it, that is find the “best way” to do it given that timeframe, where “best” is probably evaluated under some kind of metric which has nothing to do with product or user value.
Sometimes the question “explain me why it should take more than two days, when i know it takes two days to do just that in the context of a minimal MVP” goes a long way in making programmers focusing on delivering the maximum value for the product.
So I think there is actually some value in having a technical person continuously challenging developers in shortening their path to implement a feature.
I've been getting paid to write code since 1998. The last few years in management. So I know something about actually being a programmer.
I hired someone who was awesome from a technical knowledge perspective. Friendly, personable, smart, driven, etc. Loved talking tech with him and he had really great ideas.
Anyway, the problem was, I'm running a startup and every single project he was on he tried to model it as "the perfect open source project". So instead of doing something simple in a couple of days he would build this really well abstracted, over engineered (but pretty damn good code!), "beautiful" thing that would take 3 or 4 weeks to deliver.
In the end, it didn't work out for him.
Anyway, my point is that it is not true that programmers hate working on the same thing for longer than necessary. I do. You do. Many others do. But some just love being architecture/purity astronauts and refining that 3 line method into a 7 class inheritance hierarchy.
I do the opposite, I say well it's not entirely clear how much work this is going to take, I'll need to do some initial assessment to get a more accurate idea.
If they continue to pressure I ask them what changes they imagine need to be done since they're so assured of the scope and timeline of their request. I have not a care in the world of letting someone else be the "expert." If they are, I'm more than happy to go with their idea.
I've yet to get a response back outside of a collaboratige conversation with another developer working on the same codebase for similar efforts. This is usually where people realize just how much ignorance they have around their request and that they should leave it to the people more familiar with the work to make accurate assessments.
Yeah, couple of months - me, every time.
Also, fast and slow, hard and easy - those terms don't always correlate. Easy tasks can require a lot of time while hard problems might result in a quick solution after careful consideration. Furthermore getting hard tasks done quickly might be exactly why you're billing that much. Maybe you've dealt with a similar problem before - in that case your experience and competence in that particular subject shouldn't devalue your work.
What I'm trying to say is: Don't base your price on time spent alone.
Often times asking for more information can speed things up and lead to a quicker resolution.
After asking the question I get a meeting schedules where design by committee happens. The story or bug your working on gets completely redesigned. Sometimes in an illogical way. Requiring more questions.
When working remote, like most of us right now, now I will end up waiting for a long time to get a response. I then must start some other story/task/bug. Trying to context switch multiple complex tasks at some point your productivity slows down.
I have also received criticism for asking when I could figure it out myself even when figuring it out may have taken hours more time then just asking a yes/no question.
When working as a support agent I have spent a lot of my time acting as a filter to stop making the end-users go mad with all the requests from the devs and preventing the devs from being flooded with crap reports (of course, some always slip through). That means a lot of time spent on reproducing and testing (and sometimes pointing out where the bug is in the code, even suggesting patches). But very few organisations pay for that knowledge and very few allow agents to muck around for a day with a tricky issue to make sure the devs can fix it without a major hassle.
It is very easy to ask for more information as a dev, and very often it is needed, but a lot of the time it is just pushing the cost to the end-user when a minimal amount of detective work (or even a single quick test) would do the same and save everyone a lot of hassle. I would argue that in a lot of time when we, as devs, ask for more information we do it to push the issue into the future and to save ourself a minuscule of work now, not to have a quicker resolution.
Not even reaching out (again, assuming they are internal...if they are external then there are many other considerations) is just not smart IMO.
This isn't perfect behavior. But in some situations the perfect solution isn't feasible.
I of course don't respond to everything as black and white as this, and I'm not expecting them to do the kind of troubleshooting that me or a developer could do much quicker.
A 5 minute phone call or direct message can save hours of hunting and frustration trying to replicate the bug...
This could be as simple as notes logged against an issue about experiments done, etc.
https://www.folklore.org/StoryView.py?story=Negative_2000_Li...
I wish they _would_ ask me why those two lines took two days (in my case, it might be simple burnout; been coding for too long, and longer than a more conventional career track would have prescribed). Instead, nobody much cares whether I write 2 lines of code or 2000; same difference to them, and boils down to the all-important "delivering".
There's some intellectual and creative freedom in that which I suppose folks don't have with a code-involved boss who scrutinises their commits. But the opposite--nobody scrutinising your commits--isn't all it's cracked up to be, either. I almost never have to explain why I did something a certain way to anyone, not because I'm so important and command so much distinction or recognition of my expertise, but because nobody gives a crap. :-)
2 Stars: 2kg food
3 Stars: 3kg food
“Well KFC can do it...”
???
I rather enjoy fixing bugs, particularly really hard ones. They can be fun logic puzzles that take some sleuthing to figure out and offer multiple pay outs... First time reproducing it. Figuring out the problem. Figuring out the best fix. Test case fail -> test case pass.
> I know some developers don't like having to fix bugs, and so do whatever they can to get out of it. Claiming there isn't enough is a great way to look like you're trying to help but not have to do anything.
God, this behavior has annoyed me so much at times. I've worked with a few developers that were not bad overall, but would use the slightest excuse to punt on fixing an issue they were tasked with but didn't want to track down. Regularly weaseling out of tasks like this wastes the time of multiple people and either ends up back with the original dev or gets dumped on a more responsible worker.
> Because I took the time to verify if there were other parts of the code that might be affected in similar ways.
Not looking for other places in the code that are very likely to be affected by the same issue is bafflingly common, in my experience. Although I would say that managers are much more often to blame for this behavior than the devs. Any workplace that puts less weight on fixing an issue well than on artificial metrics like number of tickets closed is incentivizing exactly this type of behavior. Why bother getting criticized for spending all day fixing a simple bug the right way when you can fix 5 different iterations of that same bug and close 5 tickets in the same amount of time?
Thankfully most places I've been at -prefer- a smaller fix even if it takes a little longer to figure out.
A lot of this depends on the environment and circumstances. If you're in the middle of working on feature X it's very annoying to have to drop it to look at a bug, sometimes it's necessary but usually it can wait. This is where having great user support comes in too, capturing what the user was doing, getting relevant logs and knowing how to reproduce are important and if you don't have a good support team that falls to the developers.
The other big factor is external pressures, if you have management asking for frequent updates and putting pressure on to get through tickets quickly (especially common at consultancy type shops) then bug fixing is miserable high pressure work that I will avoid at all costs. An environment without those and bug fixing can be fun, give me the biggest most spaghetti like enterprise system and no time pressure and it feels like getting paid to solve a giant Sudoku puzzles all day.
While we're all swapping war stories I'll share my most epic 2 line fix at a place were I was afforded the time. I was working on this huge mess of over abstracted, multi threaded, spaghetti enterprise OO, trying to track down a bug that happened maybe once a fortnight. I tried narrowing it down to reproduce but nothing was working, the stack trace was about 15 levels of indirection away from the trigger so the most I could narrow it down to was hundreds of thousands of lines of code. After a couple of weeks of trying things and getting nowhere I told the boss we'd probably never track this down but they insisted I keep trying. Eventually I wrote a script to copy all the logs locally where I could search and do some analysis on them, after grepping 18 months of logs I noticed that on 3 or 4 occasions the same error was happening within 5 seconds of each other. From there it was a matter of finding "Sleep(5000)" in the code to know exactly where the error was. Turns out that 15 levels of indirection was quite slow and getting a stale piece of data we already had anyway so the time wasted turned into a nice little improvement.
The scripts for the logs become invaluable too. So many times we got "you incompetent idiots broke x with your last update" we could reply a minute later with "x has been happening since <time before any of us worked there>, you only just noticed".
Knowing which lines to add, and where, took 25 years of experience.
That's what you are paying.
Personal anecdote, the 'old' UI I'm working on had an issue where a dialog window's action buttons would be outside the visible area and people had to scroll all the way down. One guy they hired part-time spent several days trying to figure it out. I came in and added a few lines of CSS and it was fixed. The bug had been open for three years. The CSS in question:
.dijitDialogPaneActionBar {
position: sticky;
bottom: 0px;
}This might be coming from a noble place but sounds a little like shooting yourself in the foot. Bugs that can be reliably reproduced are the easiest to fix, and I've found the quickest way to get to a set of reliable reproduction steps is just to ask exactly what the user was doing when the problem happened. They don't always remember, but often do. Sometimes they even remember the time, which can be really useful for digging through logs, which otherwise are too voluminous to be relevant.
Maybe it's a cultural difference. But maybe we could "show our appreciation" for the bug report by just saying so ("Thank you so much for taking the time to report this issue. Users like you play a big role in helping us improve our software"), instead of soldiering on in the dark for 2 days.
It takes a while for a lot of junior engineers to realize small elegant solutions are better, and requires good mentorship and code review to get there.
// return the maximum value from an array
return array.min();
Only users with role X can access this content.
[where X should've been Y]
Though we usually didn't run into the latter case, ours were mostly embedded systems. So the problem was more like: You implemented this against Industry Standard 1234-A, but we're implementing against Industry Standard 1234-B which says that message Foo word 15 now has a range of 1-20, not 1-19, and 20 means "cease operations", and 18 has been changed from "cease operations" to "halt and catch fire".
So one system was developed per one spec, and another per a slightly different spec. The other common scenario was that the specs weren't created from thin air, but rather based on a preexisting system. And the authors of the spec misinterpreted what the preexisting system did and gave an incorrect specification for a behavior. When testing against the old system (or with the old system as these were mostly communication systems) you'd see a difference in behavior or failure to communicate. But since tests were never truly comprehensive, many of these errors could make it out into the world.
[User inputs X1, system displays Y]
On system crash user input was X2.
Is this specification or program logic error ?
1. Should it crash? (almost always no)
2. Should it process the garbage input as though it were valid? (pushing the input validation problem further down and potentially causing issues in random locations of the program)
3. Should it reject the input and request a different input? (probably)
Once you get to 3 you've got a number of ways to re-prompt the user or indicate that the input is invalid in a way that won't crash the program. You may need to go back to the customer to figure out their preferred resolution. But crashing is almost certainly not what they want and a sign of a program error (not specification error).
I would classify it as a specification error if they told us X1 would work, and then supplied X2 in a way that was close enough to pass most validation, but not close enough to work correctly.
Like, "The data format is a series of messages. Each message consists of up to 512 16-bit words. Word 0 specifies the length of the message, including itself." and then it turns out that word 0 specifies the length excluding itself causing us to not grab enough data in the first message, and then random amounts of data after that.
Seems
more
like
a
breakdown
in
negotiating
the
contract
than
a
breakdown
in
communication.
No
customer
is
actually
interested
in
having
the
programmer
do
their
work
in
line
with
a
compensation
scheme
that
pays
more
for
more
lines
of
code.
if (false) {
/* original code stays here, as we are payed by LOC */
} else {
/* write your new code here */
}Get 80k commits in, before lunch.
This is just one of the reasons why I appreciate it when managers have at least some coding experience.
If it wasn’t caught before prod, it’s either such an edge case as to be almost impossible to catch OR (more likely) it’s representative of a systemic failure within the organisation.
No one was upset at the fix -- if anything, the checkin's brevity communicated its correctness -- and I got a couple of pats on the back for it.
The author's approach is good for external bug reports, but they don't clarify that's indeed the case here.
I have to strongly appreciate the author for finding the root cause and tackling that instead of the symptom.
So often, especially in front end coding, you will see an exception being thrown because of a null value being passed in, and the "fix" checked in by the developer basically returns the default value if null is passed in, when they should be investigating and fixing why a null was passed into the function in the first place.
If your function has a contract that forbids nulls from You resolve the immediate bug, but this almost certainly leads to multiple bugs being created in the future (or worse, something that is quietly wrong, because 1 row in a 100 row table is missing and no one noticed) until the root issue is resolved.
1. Picking a good manager is very important!
2. Communicating with your manager is very important!
The code base was full of commented out code (the best case scenario as you know immediately it can be deleted), methods which had been deprecated and replaced with ‘myMethod2()’ and eventually ‘myMethod3()’ with all of them still in the class (and for extra fun, it wasn’t always the case that all references had been updated to the newest method), and thousand line blocks of code which static analysis helped me pick up were actually not possible to ever actually execute, etc.
Basically a static analyzer with an analysis mode for finding dead code just flagged pretty much the entire codebase.
And in the process I split a totally unrelated project which had been grown inside that codebase like a tumour into its own codebase.
Needless to say two weeks of my time invested really pepped up development velocity for that team.
Or the real reason it takes 2 days is that the code base is a big ball of unbuttered spaghetti with no tests.
I was lucky enough to be able to rewrite a couple of products from scratch with the benefit of hindsight. When the code is loosely coupled and well organized, it's rare for any of the reasons listed in the article to stall development. When the code base evolved unpredictably over a couple decades, the article is spot on.
(issue deep in the storage stack with exotic hardware only available at a customer with all debugging going ping-pong over an issue tracker between me and a customer engineer)
I think the OP covers this case very well.
> Some developers would have immediately gone back to the person reporting the problem and required more information before investigating. I try and do as much as I can with the information provided.
Some developers have a hard time with interpersonal communication but you can't isolate yourself if you're working in an org. That mindset will inevitably make you less effective (I learned that the hard way).
If this is hard for them to understand, the confused look when you reveal a change was a net removal of lines/statements must be amusing!
> Because the issue was reported with a vague description of how to recreate it
This is something my current management fully understands, but I wish we could get through to our clients. Short of being actively rude about bad requests I've run out of ideas over the years. Luckily these days we are big enough that I usually have a bit of a shield (provided first line support people and BAs who I trust) between me and direct client contact.
I'm quick to put tickets on hold as "needs more information", and been around long enough that I've developed the confidence to respond to "this is urgent, there is an SLA" with "and that service level agreement covers a minimum level of reporting from you before even an urgent matter can be progressed" or more facetiously "then it is urgent that you furnish me with the information requested", but while they accept it each time they never learn to give better details up-front next time - we still get reports of "an error" or something "not working" or, even worse, the open-ended question "is there a problem with...?".
Obviously this can't be applied to truly urgent issues, but they are usually massive problems that we are already aware of and working on before the client contacts us because, for instance, we've had an alert that something is down (sometimes we tell the client about an issue and resolution ETA before they notice, which they seem pleasantly surprised by and thankful of).
" - I have a problem, the platform is broken" " - Sure, do you have any rough idea when that happened " (I discovered that weirdly, people are very good to report a bug a day or even more after it actually happened) " - Around [some hours range]" (Get the logs for this time and this user) " - Ok, I see every information I need in the logs, it will be fixed soon"
And that's pretty much it. Oftentimes, I don't even need the time bit, just finding the user history in the logs is enough.
I'm always puzzled at engineers rambling about "users never learn how to do a correct bug report", but it seems like they themselves never actively learned that and integrated it in their everyday life.
Since users indeed never learn, stop expecting them to !
I don't expect them to. That doesn't mean I can't be irritated that they don't!
We have various logs and audits too, that can usually be used to work out what has happened. Sometimes alerts based on those logs mean we are already fixing the issue before any user reports it. Even with full audits and other logs, it is usually still quicker to locate and diagnose a problem if given what useful information we know the user has access to (the standard set: approximate time (at least the date if not today), what were you asking it to do, what did it do instead, any error messages they were displayed).
If the user is give a message with a code that explicitly says "report this number when you report the issue" and they don't bother (they just say "I got an error") and that information would save me time, you bet your arse I'm pushing the issue back onto the queue and getting on with some interesting dev/infrastructure work until a better report arrives, or looking at another issue where there is decent information.
Want to be at the head of my TODO list? Then make at least a minute amount of effort to help me help you.
It particularly bothers me when clients who negotiate a discount because they won't need first-line support as they are "big enough to have a department that triages that sort of thing", still send through terrible reports because their idea of triage is just hitting the forward button whenever an email comes in. I've got better things to be getting on with than providing free outsourced work for a bad IT department!
I'm not very customer facing these days, since we have grown to the point where I have a bit of a shield provided by our support/PS/BA teams, which is good for both my irritation levels and the clients!
this reminded me quite a bit of this deck -- in particular, a focus on a shallow metric (in this case LoC) as a proxy for measuring complexity.
Does that help?
1- Copy-paste an entire class into a new class and change a single constant in it, because he was too lazy to do inheritance.
2- "Solve" multiple bugs a day that he had introduced himself the day before.
3- Loudly complain about other people's frameworks/codes.
He was the super confident type even though he was wrong more often than not. But paired with a non-tech lead with his own impostor syndrome, it was a recipe for disaster.
It takes more work to produce succinct code and it can take surprising amount of effort and care to land at a simple solution to a complex problem.
My solution is to try to involve other people in my "process". This helps me transfer some knowledge of the decision process, helps debug ideas early on and hopefully is useful mentoring for the team.
I can do this because I am senior engineer / tech lead at my org. For other engineers I highly recommend pair programming and constantly rotate developer pairs so that everybody can get some appreciation of everybody else.
I remember working on a bug and writing tonnes of unit tests, a bunch of scripts and eventually implementing an entire e2e suite when it wasn't reproducable anywhere else.
Change to actual shipped code: 5 chars.
Users who work with IT often tend to give better descriptions and test cases. Quite often I need more information though. He's right that you try not to bother the reporting user. Sometimes there's no other way.
Reproducing a bug is often the most time-consuming part of a bug fix. It's doubly difficult when you have a shared test environment and the bug leads you into a shared data set. For instance, we have a scheduling table that's used by many applications. I can't change the data, even on test, because it can easily mess up other teams. So I have to make a copy of it to my schema, alter data, and point the code to the altered copy.
"If some code is throwing an error, you could just wrap it in a try..catch statement and suppress the error."
Yes, these developers exist. I worked with several over my career. They are frustrating because they leave damage for other people to fix, often at the worst time.
"Finding the exact cause of a problem, and looking at all the ways to get there can provide valuable insights."
Yes a thousand times. Bug fixes are opportunities to learn more about a system and, often, the user area for whom we wrote the software.
"I want a fix that isn't likely to cause confusion or other problems in the future."
A good fix takes into account the overarching software design and fits it if possible.
"I don't want a bug to be found in the future and for me to have to come back to this code when I've mentally moved on. Context switching is expensive and frustrating."
Writing software is building an abstract machine in your mind. These machines get complicated. Even when you're fixing a system written by someone else, you need time to "load" the machine into your mind.
The only time I don't like fixing bugs is when I'm against a deadline on writing/modifying another system. The context switching is a deadline killer. But, it happens. Nowadays I let the project manager know that I'm switching to a bug fix, give an estimate of how long I'll be away from his/her project, and my best guess on whether or not it'll cause a deadline slip.
Great post.
This single change took more than a week of debugging.
Outraged at having to pay $100 for two clicks the customer demanded an itemised bill. The engineer wrote:
- 2 clicks: $0.05 / click
- knowing where to click: $999.90
https://www.theguardian.com/info/2019/dec/02/faster-postgres...
Sadly, there are many projects out there which obviously failed to reach this target at some point.
For most jobs they don't do themselves people tend to understand the time it takes.
Had to explain that that project was held together with duct tape, not one bit of it was reusable, and even the engineers who build it were saying it was shit.
Adding two lines: $ 1.
Knowing which two lines to add: $10,000.
Doesn't this imply that the'd be more than two lines of code unless they're not counting the tests.
So:
1) user wants X done. Focus on that. Help the user X done. First iteration you have something crappy, doesn’t look great, too simple. That’s okay. Did it get X done. Cool. If you shipped some pretty UI but users don’t actually use it since it doesn’t solve the problem. That’s not productivity.
2) now that there are users using the tool, watch them use it. Ask questions, look at th me dashboard. Optimize their flow to do X. Make the UI delightful. We know with good certainty that this is a decent solution. Make it fast. Add tests. Harden it up so someone else doesn’t break it accidentally.
So if someone added two lines of using some existing node module that solves the user’s problem, those are very productive two lines.
> If some code is throwing an error, you could just wrap it in a try..catch statement and suppress the error. No error, no problem. Right? Sorry, for me, making the problem invisible isn't the same as fixing it. "Swallowing" an error can easily lead to other unexpected side-effects. I don't want to have to deal with them at a point in the future.
I actually did work with someone else who did this sort of thing, however it is certainly not the normal in my experience, not for anyone who takes any pride in their work.
Including this one really devalues the "it took so long because I'm so professional" message IMO.
They had a shitty bug. It took me a month to find and fix in the code. Wrote about 5 LoC. Nobody asked how I fixed it, they just were happy that I did.
IMO the real problem is trying to evaluate an employee's productivity on the scale of two days. There isn't enough context to understand the situation.
I mean I get where you're coming from, if your job is basic data wrangling (webapp -> rest API -> back-end -> database and vice-versa) then you don't need to put too much thought into it and just need the output. But that's only one part of software development.
Everyone in the thread is defending the OP because the issue is relatable. "He's productive, his manager doesn't understand!" Ok, maybe. But also maybe his productivity is way below what he's being paid for. We don't know, we don't have enough context to judge.
Snark aside, it's quite a good list.
a classic.