Log file Viewer

[10:08:19] <McSinyx[m]> hi, regarding GSoC, I want to work on parallel download for installation, which possibly involve the on-going dependency refactoring

[10:09:51] <McSinyx[m]> I am unsure about the scope of this, but very likely I'd need a mentor, but I want to do it regardless if it fits into GSoC constraints

[14:53:28] <pradyunsg> McSinyx[m]: Oooh.

[14:53:49] <McSinyx[m]> is there anything wrong withmy plan

[14:54:19] <pradyunsg> McSinyx[m]: this would be great to do, but I imagine that would probably be a little too big for GSoC, in my opinion.

[14:55:52] <McSinyx[m]> are you referring to the parallel or the dependency thing

[14:56:01] <pradyunsg> I mean, I say that as someone who has been working on my GSoC project for the past 3 years, so I'm genuinely not sure if I'd call it a disqualifying characteristic. 🤷🏻‍♂️

[14:56:23] <pradyunsg> McSinyx[m]: I think both independently are fairly no-trivial tasks.

[14:56:46] <pradyunsg> Together, they compound each other's complexity.

[14:56:49] <McSinyx[m]> I can recall that the dependency resolving part is your GSoC 2017 lol

[14:57:02] <McSinyx[m]> but they must be solved together

[14:57:22] <pradyunsg> Not necessarily together, IMO.

[14:57:28] <McSinyx[m]> or at least we can't download and build in parallel without proper dep. solve

[14:57:36] <pradyunsg> Exactly.

[14:58:02] <pradyunsg> And we're not going to have the solver done until the end of the year, at least.

[14:58:14] <pradyunsg> *closer to the end of the year.

[14:58:38] <McSinyx[m]> what is the current approach, the project board is a bit scatter for new comers to scope I'm sorry

[14:59:02] <pradyunsg> Current approach for?

[14:59:02] <McSinyx[m]> also do we have any plan to drop py27 anytime soon?

[14:59:12] <McSinyx[m]> for dependency solving?

[15:00:19] <pradyunsg> Re py27: not yet. I'm pretty sure we all want to, but we don't want to drop it and leave too many users hanging (pip is one of the ways people dealing w/ Py2/3 stuff get their backports).

[15:01:28] <pradyunsg> re dependency solving: we're implementing a resolvelib-based resolver, by implementing all of the pip-specific quirks in pip's codebase and the resolution algorithm stuff in a separate codebase (resolvelib).

[15:02:49] <McSinyx[m]> thanks for the py27 info

[15:04:52] <McSinyx[m]> I think I'd need sometime to read more about resolvelib and get back latter on

[15:05:21] <pradyunsg> Sure, go ahead.

[15:06:16] <McSinyx[m]> my plan is to implement something analogous to aptitude logic, which has been quite a good experience for me dealing with dep. hell

[15:07:54] <McSinyx[m]> anyway I'm on the matrix bridge so you guys can be sure that I won't miss any message (plus this channel is logged); see you later

[15:08:16] <pradyunsg> heyguys.cc

[15:08:54] <pradyunsg> I encourage you to build an understanding of the situation before taking a stab at the problem, and write down some sort of a broad plan.

[15:09:35] <pradyunsg> Maybe you already have that understanding, in which case writing down the broad plan can help communicate that. 🙃

[15:09:57] <pradyunsg> !logs

[15:09:57] <pmxbot> http://kafka.dcpython.org/channel/pypa-dev

[15:21:14] <McSinyx[m]> thank you for your advice, I'll try to organize my thoughts while diving in