Show HN: CodeMill – a marketplace for pull requests | Hacker News Reader

Show HN: CodeMill – a marketplace for pull requests

codemill.io

169 pointsshaharsol80 comments

BountySource (https://www.bountysource.com/) is a similar thing which is well established and doesn't mess up my browser's scrolling.

Having sane browser-scrolling should be HN 101 submission etiquette.

I read this a lot but I don't get it. Can you explain to me what the problem is? I scrolled the site both with my touchpad (2-finger gesture) and with the touchscreen and it just, well... scrolled. What is wrong here?

It doesn't scroll in the same way to other pages on your browser.

It's some bizarre trend these days where people feel the need to customise how the scroll behaviour works.

Perhaps HN could implement this. I'm not sure we can ever expect submitters to test a site's scrolling functionality out of all the browsers out there or even say 3 of them. This site for example is working just fine for me in Chrome.

There is nothing for HN to implement. The problem is not that scrolling needs to be tested and fixed in multiple browsers. The problem is that this site uses a JS plugin to modify the browser's default scrolling behaviour, and this plugin provides a poor experience for some users (myself included).

Agreed. My Macbook Pro with Chrome scrolls like shit on this.

Too slow at the top, then too fast in the middle, then too slow again further down.

I'd tweet to them but I don't see an account on Twitter: https://twitter.com/search?f=users&vertical=default&q=codemi...

Chrome 49 / Linux: scroll wheel doesn't work, page up/down keys don't work.

It's using an old (v0.9.9) version of smoothscroll plugin. This issue was actually raised in its Github Repo [1]. Reportedly new version detects if the input device does inertial scrolling and disables itself in that case.

Of course a better solution would be to just remove the plugin ;).

[1] https://github.com/galambalazs/smoothscroll/issues/85

Does BountySource give bounties for itself?

All screening costs are incurred by the developer - that is, they have to wade through every CodeMill issue (wildly lowball $5 offers included) to find one that's worth their time. They then fork and study the codebase, performing the work on spec without ever meeting the organization or reaching a formal agreement. And then ... hope they get paid 90% of the offer?

This arrangement sounds remarkably unfriendly to developers. Expect high churn among experienced devs, if they even show up at all.

The custom scrolling on the site bothers me a bit. Its really cumbersome when you are using a trackpad, and it does not scroll the way you expect to. It also prevents me using left-swipe to go back a page.

Even with a scrolly-heely mouse it sucks. My OS has a perfectly good scrolling mechanism, stop messing with it! Especially when the result is worse!

hehe sorry for that. You know I used a ready made theme, and it works fine on my chrome/mac and I didn't really tested it on many other OS/browsers. What you're using?

funny that you mention chrome/mac, because thats exactly what I am using.

chrome:Version 48.0.2564.116 (64-bit) osx: 10.10.5 (14F27)

I guess it's something in the parallax the theme uses then. I'll need to have a look. What puzzles me is that I don't really feel any problem myself...

Well.. smoothscroll.js

This removes it if you wanted to A/B test it

    window.removeEventListener("mousewheel", ssc_wheel)

I don't find it that bad but it's a bit more janky than default. But you lose the overscrolling that OS X has.

Thanks!

It's awful beyond words, you should never ever mess with the user's scroll. If themes have it just delete the feature entirely.

I don't see any issues at all on Firefox on Linux.

I suggest you use codemill to get a dev to fix it for you .

Agree with the other comments. Love the site. Hate the scroll.

Is there any extensions which would disable features like this?

Use a 3rd-party blocker and block smoothscroll.js as that's the culprit 95% of the time.

I have this in my uBlock filters:

    !Disable Smoothscroll
    */smoothscroll.js

Interesting, but they should make it more clear how does it work legally - i.e. are there guarantees built in that the repo owner won't discard the pull request, and then use the code anyway? Especially on open-source projects. Is this purely reputation based? Seems like it could be done with proper licensing. Also, the line

  >If you're not happy with their progress though, you can unassgin them
 and then other developers will be able to grab it again.

Worries me a little. Is there no limit, no appeal process? you could be 80% done with a task, and then the repo owner decides they don't want it done anymore, and just ditch you?

The idea here, since we don't want to make it a competition or something where only one wins the jackpot, that if a lazy dev takes the job, and he's behind schedule, or provides a low quality deliverance, then the client can unassign. I here your concerns but I hope it won't grow into a backdoor for stealing code.

Builds backdoor for stealing code... hopes it won't grow into a backdoor for stealing code.

Not entirely fair. A lot of good things have worked on the honor system, as long as fair thought has gone into the "rules" of the system so that enough easy, edge case abuse scenarios are protected against. A big one that comes to my mind was the gaming league, CAL, back in the day. It was 100% honor system based, with a target audience that was known to be hackers, liars, and in general kids - yet it worked!

Paying upfront also implements the honor system.

So (effectual or not) attempts to protect consumers from developers not delivering. What symmetric mechanisms do you have in mind to protect developers? How are you handling copyright assignment, which should really wait until payment has cleared?

For FLOSS you could let users put donations on an issue, whenever the pot gets large enough for a developer to take interest they'll work on it. That way you don't need to determine a price and you let users decide what features they want by putting their money where their mouth is.

nice idea.

> let users decide what features they want by putting their money where their mouth is

While nice in theory, something like this needs to also have the FLOSS project developers mark requests as feasible and/or within the scope of the project's goals before allowing bounties to be placed on them. Otherwise there is a risk of project hijacking.

Yes, chrome extension so that next to the +1 button is a way to chip in money, and adds a comment to the issue like "seibelj pledged $10 to this issue".

Like Bountysource?

It takes a lot of time to get up to speed on an unfamiliar codebase. I signed up and currently see two jobs available - one for $5 and another for $7. Either the offers will have to be more compelling or you need a lot of PR's on one project to make it worth it for developers. You could focus on stand alone projects that don't need to be integrated at all, but then you may as well do that on a rent a coder type site instead of Github.

Typo:

> you can unassgin them

Anyway, this seems interesting. But as a developer who completes pull requests, what's to make sure I actually get paid? The buyer could easily reject a pull request and merge the code elsewhere anyway. You say the payment is pre-approved. Does that mean it's impossible for the original developer to get their money back?

Plus, the repo owner can change the price at any time, from what I gather.

You're right. I didn't think of that. Need to make sure payout doesn't change while task is grabbed.

Thanks for the typo fix.

It's true, a client can do what you describe but then we won't keep them as clients for long ;-) I assume in the future we can add a mechanism to ensure payment in case of such a code "theft", and I admit this isn't handled now. That said, we're talking rather small tasks, not complete projects here, and what you describe is not something that doesn't happen in real life, i.e. contractors that don't get paid. I assume that a developer that works on this system, saves a lot of time on marketing and clients relation and even if 1/100 tasks he is not paid for, it still be worthwhile than the alternatives.

Isn't a parallel (and perhaps even larger?) market that of users of a system or library that might be willing to put a bounty on having a feature prioritized by the maintainer(s)?

With that in mind, could you use a variant of the same system to put a bounty on issues/tickets being resolved?

I don't know how to (in general) solve the problem of developers at large enterprises taking advantage of open source developers that are barely making ends meet. Nor the problem of developers being demanded to fix bugs or add features for the benefit of people who aren't paying anything for it -- but maybe this could be a start?

If I understand what you're saying, we made an open source offering based on our current model: http://codemill.io/for-open-source is that what you're after?

I don't like the name. It immediately makes me think "diploma mill," which signifies cheapness and inauthenticity.

You know, the original name we came up with was GigHub. We also found the owner of the domain gighub.io and agreed to buy it. But then I thought we're a GitHub app and the names are too similar and maybe github won't like that, so I asked them, and yes, they said it'll violate their trademark policy and we better find another name...

Great names are made not found. Don't worry. If your service is great your name will be great :)

This seem like a good idea but in practise i am not sure how this will work for large codebase where more time is spent on understanding and guessing where to make a modification or adding a feature than writing code.

And how this will not be flooded by devs from poorest countries of the world like upwork and a lot of freelancing websites making those platforms pretty much useless for westerners.

I believe that if you can define your task in as much details, and break it to the most micro level, then it works. If on the other hand, you're issue is something like "there's a bug when sometimes an invoice is inserted with a negative amount", so yeah, this is something more comprehensive, and maybe related to god knows what in the codebase....

Went to website, noticed scroll override, closed site. If you're inexperienced enough to think scroll overrides are ever a good idea, your product probably sucks way more.

I see a bit of a conflict between

    Does CodeMill work on private repositories as well?
    Yes, the only difference is that interested developers won't be able to fork them unless you authorize them...

    How do I screen developers?
    That's part of the beauty of CodeMill -- It's based completely on GitHub, where each developer has their public profile

This might be useful for orgs sponsoring open source projects but for orgs trying to get cheap labour on internal projects, there's no way of appraising previous similar work.

(Also, I have only ever heard the word 'mill' is used in a pejorative context: "Diploma mill", "essay mill", etc)

I agree. I assume however that on average there's enough open source pushed for each developer on github. Didn't see any stats or whatever, just a hunch feeling I got from years of using this platform.

As for the name... I replied to another comment. Please ctrl+f "GigHub" on this page to find it :)

If only the pull requests were the only part that would a payment incentive. How do you encourage people to do code reviews if they're only paid for pull requests ? How about testing, bug reporting and triaging ?

I here what you're saying. All of these are also tasks that need to be done. however I believe that in most scenarios, those tasks would have to be held at the organisation level. I'm also not sure the github platform integration would be useful to accomplish that, and the coding itself is the basis for the dev process, we thought we can help with that.

I understand. Let me rephrase that. What you provide is nice, has the easy feeling of providing some value, but is IMO typical of many startups. You just used whatever API was available to you, and used them to build a product on top that some people may want, but does not solve a hard problem. There exist competitors (gittip, code bounties), albeit not as integrated with Github.

Solving a hard problem in this case would be doing the things that don't scale (http://paulgraham.com/ds.html ). That is, providing the tools to oil the development process by incentivizing the boring yet essential tasks that I talked about (QA, triaging, etc.).

And I don't mean that you provide no value. It's indeed quite an interesting proposition. I just don't see it going further than a side project you'll have to maintain forever for a small revenue on top of your current job. Which is perfectly fine, I have a few of those myself.

I signed up, but Open Tasks is empty right now. It might be interesting to report info about completed tasks---to give an idea of the prices and type of PRs being requested.

This is an interesting idea with several possibility of incentives: project owner, project users, or companies employers could sponsor these PRs.

Also could be useful for learning/tutoring---formulate a question you're stuck on as an Issue and let an expert show you the way with an educational PR.

Many tasks had been grabbed in the last 2 hours since this post is on HN front page, however I still see several available tasks. Do you see it totally empty (which implies a bug) or by empty do you mean "just a few"?

Mine is empty = zero tasks as well.

Oh, I see. I bet you're looking at the "open tasks" while in the "clients dashboard"... to grab tasks and work on them you need to get to the "developers" perspective, which is done by hovering on your profile image on the top-left and a menu will appear ...

I'd suggest making that more obvious somehow. Before seeing this comment, I had no idea how to see open tasks that are looking for developers to take.

Same problem here. Sign-in with github was very nice. Then drops me into the dashboard. Open tasks are 0. Its a UX dead end.

Interesting. Cool idea. Thanks for the post!

Just guessing, but wouldn't it work better as a 2-phase process? Phase 1 you set a price on automated acceptance tests. Phase 2 developers bid on how much to write code to pass the tests. That, and perhaps required paired-programming, should address the quality issues you're probably going to face just setting a price on code slinging.

This is something worth thinking about. Of course time, and experience, will tell. We wanted to keep it as simple as possible, and as disruptive as possible to the current employment/contracting process that goes "offline", so we decided to start the way we did. We're not shutting our ears though for ideas.

Writing good automated acceptance tests for something the size of a PR amounts to 80% of the work.

I really like general idea. I was thinking along the same lines. Essentially by splitting your work well, the skill that I have as experienced dev, you can get a lot of people working on it. It does require discipline from all involved.

Tasks available are not tempting at all, but I guess if you can get good devs to be present, this can became platform to do work.

I am not huge fan of the interface etc, but this is start, so this is OK.

ugh. as someone who wandered over to development after years of freelance design, this brings back terrible memories. you may want to check out http://www.nospec.com/ and spend some time ruminating on the ethical ramifications of the service you're offering.

I appreciate explaining exactly why you need permissions, but being asked for each and every permission individually gets tedious.

I totally hear ya. But then would you sign up if it had a huge list of permissions to begin with? I'm really curious, not just rhetorical.

Frankly? That's not really I question I should be answering. That's a question you should be a/b testing.

For example: Option 1: sign up with all permisions Option 2: sign up with 5-6 most common permission, request additional for least common Option 3: sign up without permissions, request every permission as needed.

Then check what matches your conversion goals.

Come on.

Asking for advice in a startup forum from someone who was clearly opinionated and this is what you come up with?

Answer the question and then, -for extra credit, recommend A/B-testing.

@shaharsol: not every piece of advice is good advice. That said A/B testing can do wonders but don't waste too much time. Deploy it well in advance of next marketing move. (e.g hn link submission etc)

There's a clear user preference question here. Do users prefer being asked upfront for all the permissions at once, or do they prefer being asked for permissions on the go.

Random dude in startup forum isn't smarter than your users, and can't anticipate how your users really would behave.

The email permission confused me, seemed like it was a choice I had. But since there were no other buttons there I assume you can't use the service without allowing it.

10% commission taken out of each task price? That's steep!

see upwork and head hunter fees. Also groupon et al. 10% is cheap/norm.

I amazed how many sites, dedicated for developers have scroll hijack. I see such website at least once every every day on the front page of hacker news.

That's because we can code but can't design. So we take ready made themes and we don't pay attention that they have scrolling problems...

What about https://snowdrift.coop/?

Have you considered an explicit assign? Seems like it might make the process a little saner.

Typo: `seamlessly` not `seemlessly`.

Unbelievable,Awesome

is it just me or is the logo blurry?

It looks awesome! However:

-------------------------------------

Do many developers work on each task? No, just one at a time. Once a developer grabs a task, it's theirs. If you're not happy with their progress though, you can unassgin them and then other developers will be able to grab it again.

-------------------------------------

Let me requote the relevant term:

"unassgin"

Where is the option that does not involve rectal juniper?

Reply on news.ycombinator.com