[Haskell-cafe] RFC: GSoC proposal: Improve the Hackage login system

[Chris Wong]

Wow, thank you for the comprehensive critique! I'll respond to your
points individually.

On Thu, Mar 20, 2014 at 11:26 PM, Mateusz Kowalczyk
<fuuzetsu at fuuzetsu.co.uk> wrote:
> I've read your new proposal and I think you'll definitely need more
> solid specification on what you aim to do. An example is that you state
> 3 things for the ‘rewrite the homepage’ aim but the goals are
> ’rewrite/reword something, move an advertisement, add some links’. While
> these are perfectly fine goals, they do not fit your timescale at all!
> Just this aim alone you schedule for nearly a month! It sounds that it
> could be done in a day instead. By the way, I don't think that moving
> the advertisement is such a high priority that it needs to go into a
> proposal: if they're funding Hackage and want their ad on the front
> page, I don't see the problem. Existing Haskellers will rarely look at
> the front page and new ones will probably see it a few times. I doubt
> anyone would click on the About without a good reason to.

Agreed. On looking further into the Hackage issues list, I found
there's a lot more to be done.

> Your specification of the search work is very lacking too. You don't say
> what's wrong with current search and how you aim to improve it. I
> believe Duncan Coutts was working on better search results by using
> better heuristics. I don't know the details but it's almost certainly
> worthwhile to contact him about it.

Sure. He seems to be the go-to guy for everything Cabal/Hackage, so I
agree it's a good idea to talk to him.

> Here are two things that I think need attention interface-wise. Note
> that even with these, you'll probably need to state more stuff:
>
> * Better report presentation: at the moment we (sometimes) can get a
> report log from Hackage about the build status of a package. At the
> moment one has to manually change the URL to then get to dumped ‘Show’
> instance of a Cabal data type. Buttons should be added and the report
> should be presented in a much nicer way. Maybe even we could have a
> small indicator whether the build worked or not on Hackage.
>
> * Hackage exposes an API that lets us work with it. You can see it at
> [1]. While it's nice to have, there's functionality there that's only
> available through the API itself when it really should also be available
> through friendlier means on the site itself. A big thing that comes to
> my mind is the documentation uploads. If you search around the web,
> you'll see that uploading own documentation for a package is quite a
> mysterious process. It'd be much easier if we had a button which will
> take us to an upload page which also contains instructions for properly
> generating the documentation. Implementing a documentation checker (for
> example to check that cross-package links aren't broken, aren't pointing
> to local things, checks that index page works, all the common mistakes)
> could be a side project.

I admit -- I didn't even know Hackage had either of these before
today, when I saw them on your blog ;). Exposing them would be great.

> Lastly, there are ~90 open bugs on the Hackage issue tracker[2]. Many of
> those are fairly easy and of pretty great benefit to Hackage users
> (a.k.a. pretty much everyone). Things like #74[3] could be implemented
> with relative ease and I feel that if you could propose to close some
> specific tickets and work on others, your proposal would be much, much
> stronger. Frankly, I think that moving the ad from the front page is
> much lower priority than many of the tickets there ;) I actually thing
> that if one were to concentrate on the tickets themselves and make (some
> of) the interface work a side goal, nearly all the tickets could be
> fixed. Of course some tickets are best fixed with interface changes so
> it's not like that's going on the back-burner.

Thanks for the suggestion -- the proposal pretty much writes itself
now, with each issue mapping to a task in the proposal.

> A friendly reminder that there are only about 32 hours to get your
> proposal in so make sure you don't miss it!

Bring it on. It's Friday here, so I don't need to sleep ;)

> Oh, I nearly forgot, did you find a mentor for this new project you're
> proposing?

Not yet -- but it's not too late to do that, right? I notice the other
Mateusz has asked on cabal-devel; I might follow him when I'm more
awake.

Chris

> [1]: http://hackage.haskell.org/api
> [2]: https://github.com/haskell/hackage-server/issues
> [3]: https://github.com/haskell/hackage-server/issues/74
> --
> Mateusz K.
> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe at haskell.org
> http://www.haskell.org/mailman/listinfo/haskell-cafe