Gentoo Archives: gentoo-project

From: Daniel Robbins <drobbins@××××××.org>
To: gentoo-project@l.g.o
Subject: [gentoo-project] Burden of proof
Date: Mon, 02 Apr 2018 06:13:46
Message-Id: CAPDOV4_SSJwHv2YuBA=Ph1dpqWboGXtKT8D2NG25Pzbi9HSf6w@mail.gmail.com
1 I am going to post some things I have seen from a Council member, and
2 people can judge whether this is appropriate. First, mgorny derails a
3 technical conversation to lecture and berate me, saying "now I see why
4 you've never returned to Gentoo", in #gentoo-portage:
5
6 [technical conversation.... ]
7
8 2018-02-24 04:48:54 [Arfrever] drobbins: See this example
9 (version-conditional exec()):
10 https://gitweb.gentoo.org/proj/portage.git/tree/pym/portage/process.py?id=39c797992bcdf8403521d8b61bb3e592135b3307#n196
11 2018-02-24 05:28:01 drobbins [Arfrever]: yes, so that's all code
12 that can now be removed.
13 2018-02-24 05:28:19 drobbins the concept is that it adds to the
14 LOC count, doesn't add any functionality
15 2018-02-24 05:28:46 drobbins it's an awesome exercise in
16 backwards compatibility that doesn't have a large benefit on a distro that
17 is moving to 3.6.
18 2018-02-24 05:56:09 --> zmedico (~quassel@gentoo/developer/zmedico)
19 has joined #gentoo-portage
20 2018-02-24 07:32:21 -- perfinio1 is now known as perfinion
21 2018-02-24 08:24:51 mgorny drobbins: are you convinced someone will
22 actually remove all the compat cruft?
23 2018-02-24 08:25:12 mgorny portage is practically unmaintained, so i'd
24 rather not make major changes like making upgrades a nightmare for the sake
25 of few patches
26 2018-02-24 08:32:44 <-- Shentino (~Shentino@unaffiliated/shentino)
27 has quit (Remote host closed the connection)
28 2018-02-24 08:33:05 --> Shentino (~Shentino@unaffiliated/shentino)
29 has joined #gentoo-portage
30 2018-02-24 08:33:41 drobbins mgorny: yes, I am convinced, and it
31 just allows a lot of compat code to be dropped.
32 2018-02-24 08:35:01 drobbins there is additional work and
33 complexity involved in adding functionality and maintaining compatibility
34 with 2.7
35 2018-02-24 08:35:23 drobbins so if you feel that portage is not
36 maintained enough, this will help reduce the work load
37 2018-02-24 08:35:49 mgorny 'helping reduce the work load' does not
38 imply the work will actually happen
39 2018-02-24 08:35:58 mgorny most important goals so far don't have much
40 py2/py3 trouble
41 2018-02-24 08:36:19 mgorny and adding extra complexity to make things
42 faster when they need to be made correct first is not helpful
43 2018-02-24 08:36:34 drobbins there is an effort to integrate
44 async functionality into portage which is hampered by 2.7 compatibility
45 2018-02-24 08:36:54 drobbins it actually reduces complexity, but
46 whatever. It seems you are convinced it's a bad idea.
47 2018-02-24 08:38:57 drobbins standardizing on the official async
48 implementation rather than using a home-grown one for compatibility with
49 python 2.7 seems like a win to me
50 2018-02-24 08:39:11 mgorny i'd like to see async working correctly
51 first
52 2018-02-24 08:39:25 drobbins what do you mean?
53 2018-02-24 08:39:36 mgorny there is the socks proxy for distcc written
54 using async, and it has serious exception handling problems
55 2018-02-24 08:39:48 mgorny or signal handling
56 2018-02-24 08:40:24 <-- Zero_Chaos
57 (~zerochaos@gentoo/developer/pentoo/zerochaos)
58 has quit (Ping timeout: 260 seconds)
59 2018-02-24 08:40:53 drobbins I don't think that is indicative of
60 the python async implementation
61
62 [mgorny loses it....]
63
64 2018-02-24 08:41:02 --> Zero_Chaos
65 (~zerochaos@gentoo/developer/pentoo/zerochaos)
66 has joined #gentoo-portage
67 2018-02-24 08:41:41 mgorny w/e
68 2018-02-24 08:41:47 mgorny portage API is a horrible pile of cheap
69 hacks
70 2018-02-24 08:41:53 mgorny that needs to be fixed to make it useful
71 2018-02-24 08:42:02 mgorny not added more hacks so that cheap hacks
72 may be a bit faster
73 2018-02-24 08:42:11 drobbins agreed
74 2018-02-24 08:42:25 drobbins that is a good reason to use an
75 actual language feature instead of a roll-your-own async implementation
76 2018-02-24 08:42:36 drobbins since one is now available.
77 2018-02-24 08:43:02 drobbins I needed to get my tree
78 regeneration going a lot faster. It now takes 65 mins instead of over 4
79 hours.
80 2018-02-24 08:43:34 drobbins I think that zmedico is trying to
81 do things not as hacks but have a strategic plan for implementing new
82 technology
83 2018-02-24 08:43:51 mgorny so you had a broken scripting, and instead
84 of fixing it you're adding hacks to portage to workaround your problem
85 2018-02-24 08:44:03 drobbins how was my scripting broken?
86 2018-02-24 08:44:07 drobbins since you know so much
87 2018-02-24 08:44:07 mgorny (protip: egencache has --jobs)
88 2018-02-24 08:44:32 drobbins protip: egencache does not function
89 when you are working on assembling trees that you haven't generated
90 metadata for yet
91 2018-02-24 08:44:47 mgorny then you should've generated the metadata
92 first
93 2018-02-24 08:44:54 drobbins and protip: don't protip me
94 2018-02-24 08:45:19 mgorny (or used trees with metadata)
95 2018-02-24 08:45:19 mgorny (or shouldn't be assembling stuff in the
96 first place)
97 2018-02-24 08:45:19 drobbins oh yes, you are right.
98 2018-02-24 08:45:28 drobbins thanks for the correction ;)
99 2018-02-24 08:45:55 mgorny the multi-repo model has implemented for a
100 reason
101 2018-02-24 08:45:56 drobbins anything else you'd like to correct
102 me on?
103 2018-02-24 08:46:49 mgorny your attitude
104 2018-02-24 08:47:04 drobbins what is wrong with my attitude?
105 2018-02-24 08:47:15 mgorny now i see why you've never really returned
106 to gentoo
107 2018-02-24 08:47:21 * mgorny out
108 2018-02-24 08:47:23 <-- mgorny (~quassel@gentoo/developer/mgorny)
109 has left #gentoo-portage ("http://quassel-irc.org - Chat comfortably.
110 Anywhere.")
111 2018-02-24 08:48:58 drobbins I agree, mgorny out.
112 2018-02-24 08:49:46 drobbins why does he get so pissed all the
113 time?
114 2018-02-24 08:51:35 zmedico that's a good question
115 2018-02-24 08:52:07 slyfox_ i don't think it's you specifically
116 2018-02-24 08:53:56 zmedico psychology is a complex topic

Replies

Subject Author
[gentoo-project] Re: Burden of proof Daniel Robbins <drobbins@××××××.org>