Hacker News Reader

Comment by oliwarner | Hacker News Reader

There are some outrageous time estimates in this. 7 days to create an auth system with the big 3 social integrations? 6 days for private messaging? 2 days to dump a prefab social sharing blob in your template?! I'd do all of that in a morning.

I'd just be careful if you're considering using this to quote your jobs. If somebody working for me quoted these sorts of numbers, I'd fire them on the spot. I wouldn't even give them a second chance. It would clearly demonstrate you didn't know what you were talking about.

View on HN

You could do it all in a morning if you didn't do any of the following:

- Create and implement a high fidelity design.

- Include the project-specific requirements that every feature has... the auth system needs to work via ajax, connect with a legacy db, and include animated transitions. Private messaging needs responsive design and realtime updates. Etc. etc. etc.

- Deal with scope creep, clarifications of the specs, and hours of related discussions.

- Write automated tests.

- Do live QA.

- Fix inevitable bugs, browser and device inconsistencies, ux problems, etc. etc. etc.

- System administration and deployment.

- Load testing, caching, performance tuning.

If a client tried to tell me I should be able to do all this in a morning, I'd fire THEM on the spot.

You're not reading what I said. The design is accounted for elsewhere in the calculations (that's actually a bit closer to the mark) and that includes many of the things you're talking about above.

My "morning", is literally plugging django-social-auth, a custom User Model, a private messaging Model and few views to make it all work.

I'll grant you that testing might eat up the afternoon, but these are all solved problems that ship in libraries, many of which include their own tests.

Perhaps the people downvoting need to change how they build stuff.

The result of your single morning would be a clunky ux that no one wants to use (or pay for). No one's saying you can't get a good start on the data model and basic skeleton in a single morning, but only someone who's never produced working, polished software would contend a finished and tested feature of any significant complexity at all could be truly finished in a morning.

The people downvoting you live in reality...

Well, it sounds to me like you haven't worked on systems of any complexity. Seven days for a federated auth system that behaves the way people actually want it to behave is very, very low.

What's your hourly rate? I'd consider paying you to build this, to prove that you're wrong (or that you're a badass).

ok. How much will you charge for such a morning?

Then go right ahead and do it.

And when you're behind, because you keep underestimating difficulty, and the client is tired of your excuses, one of us will swoop in and show them how a real developer works.

Reply on news.ycombinator.com