1 |
On Thu, Jan 02, 2014 at 09:26:34, Duncan wrote: |
2 |
> Are you still interested in doing something with this general ebuild |
3 |
> patching framework? |
4 |
|
5 |
Hell yeah :) Was about to start work on it next week or two. |
6 |
|
7 |
> While the semantic-desktop use is gone, I still find |
8 |
> it useful for the occasional ebuild patch, and keeping the framework |
9 |
> around and working means the next time something like this comes up, |
10 |
> it'll be far easier to deal with. |
11 |
|
12 |
Well it's not quite gone, as konversation-1.4 and 1.5 pull in akonadi |
13 |
for "address-book integration"; I've patched ebuilds in my local overlay |
14 |
just a couple of days ago, following Chiitoo's post[1] (he updated the |
15 |
patch for 1.5) but not installed either yet. I've had a couple of busy |
16 |
months too, and not done much on update, but got back to it a week or |
17 |
two ago, starting with a bug that was weird (only showed up on end-user |
18 |
system, and I'm still not sure why it happened which bothers me, but |
19 |
reverting the old, minor change has fixed it.) |
20 |
|
21 |
I also want to add some of creaker's work to kde-lean, so we have a nice |
22 |
out-of-the-box experience for people who aren't interested in nubkit[2] |
23 |
either. Chiitoo's post is in his thread, and creaker's also coded up |
24 |
a Qt automounter that works in the Dolphin context menu[3]. |
25 |
|
26 |
I switched back to using Konqui for file-management: can't believe |
27 |
I was so dumb as to keep trying to use Dolphin. Back in the day, a |
28 |
KDE desktop with konqui and fish://, with just kwrite for editing, |
29 |
was considered THE best php "IDE". Midnight-commander mode *rocks* |
30 |
for work, and now I just have the Konqui "profiles" button in |
31 |
taskbar, with Bookmarks for quick web access. Much nicer :-) |
32 |
|
33 |
> And if you have a different list to take the discussion to, I'm open to |
34 |
> that too. I prefer not to go just personal mail, however, because that's |
35 |
> too easy to let drop and never come back to if something dumps on me like |
36 |
> those mountains of overtime did. IOW, were this thread private mail, I'd |
37 |
> have never done this followup... |
38 |
|
39 |
Yeah agreed, although as you may have realised I'm not a regular on this |
40 |
list: I only opened it today out of curiousity as to why it was in my |
41 |
subscribed newsgroups. Heh forgetful old me, I'm afraid :) |
42 |
|
43 |
So ok, I've remembered now heh and will try to stay up with it. But |
44 |
really, I would have bcc'ed as well: it's fine to keep list discussion |
45 |
to the list, and I entirely agree with that. But you have to bear in |
46 |
mind that you're working with human beings who are forgetful, too, |
47 |
and I for one read gentoo lists (apart from pro-audio) via gmane. |
48 |
|
49 |
[I haven't bcc:ed you on this as the address I have for you is |
50 |
clearly a list one, so I'm presuming you do everything via email |
51 |
and don't want two copies.] |
52 |
|
53 |
In terms of keeping it in public, I'd also recommend using a bug |
54 |
tracker. It's amazing how motivating it is to fix and close a bug, or |
55 |
at least I find it so. Keeping them active indicates that you've not |
56 |
dismissed that concern, even if you've not worked on it for ages. I |
57 |
have a couple that simply have to wait til I've got more fundamental |
58 |
things working, but which will be sorted out when I get there. I tend |
59 |
not to close those LATER (I think i have one bug in that state) as |
60 |
I want the reminder when I use the interface; they just get milestoned |
61 |
against the next version (though we've been on 0.1.4.0_{pre,rc}N for |
62 |
about 5 years or sth: once --toolchain is correct, we'll move up.) |
63 |
|
64 |
> At least with the public threads, there's some reason to come |
65 |
> back and wrap up, if nothing else, and that could ultimately |
66 |
> mean the difference between this project going public |
67 |
> and it staying just a bunch of scripts I use myself, locally. |
68 |
|
69 |
Well tbh, I was just going to do it anyhow, as Neddy's started a |
70 |
gentoo-static overlay[4], and I need to maintain kde-lean come what |
71 |
may. (there's nothing in atm, as i was waiting on you before, |
72 |
which was I'd decided just to go ahead with our own thing. good |
73 |
job I did open this list, huh?:) Gentoo-static is nice and small |
74 |
so makes a good test-case. |
75 |
|
76 |
I mean you say it's better to keep it public and I agree; but there's |
77 |
also something to be said for keeping in touch, and not assuming your |
78 |
workflow and methods are in use by others. For instance I'd basically |
79 |
given up on you; removed your repo from accessibility and was about |
80 |
to delete the trac and the backend repo. A further 2 months has |
81 |
gone by, imo simply because you cba to email me. Good show *cough* |
82 |
|
83 |
So the idea is to have a list of ebuilds per overlay, and simply to |
84 |
do a diff from the upstream version in-tree, which is our patch. |
85 |
Then we a) check the current ebuild version hasn't changed, |
86 |
and: b) try to apply it to later incoming versions |
87 |
(after a sync.) |
88 |
|
89 |
Ideally we want to do it _before_ q and eix post-sync. |
90 |
eix isn't an issue for our users, in that update runs it |
91 |
separately, but it may be for people who use emerge directly, |
92 |
since istr that eix-sync is used to run emerge --sync. |
93 |
|
94 |
Shouldn't be a problem with a suitable postsync hook, which |
95 |
would cover both cases. OFC it's only needed for the overlay |
96 |
author running upatch (new name in my head, easier to type) |
97 |
to maintain the overlay, not for the overlay users who just |
98 |
use layman. layman sync is normally run by eix-sync iirc, |
99 |
which is perfect if we've modded the ebuilds just before. |
100 |
|
101 |
However it's important to know the version of the upstream ebuild |
102 |
so we can track changes (ie SHA256 or MD5 from manifest.) I'm |
103 |
leaning toward simply backing up the manifest file, as that |
104 |
covers everything. |
105 |
|
106 |
If this is sounding a lot like a git branch, that's because it is. |
107 |
I've looked around for a suitable gentoo git repo, but can't |
108 |
find one (perhaps I haven't looked hard enough.) There isn't one on |
109 |
gentoo's github[5], for sure. Funtoo used to maintain one, but |
110 |
they're not any more, and I can't see any branches in their repos |
111 |
online that are to do with that. |
112 |
|
113 |
Patrick started the conversion process, and left the intermediate |
114 |
stage work-products up[6] for anyone who wants to continue it, |
115 |
although it would ofc be much better if he also showed how to |
116 |
update it with current changes. It still saves a lot of time |
117 |
though, as you need a massive amount of RAM (for me anyhow) to |
118 |
do the initial conversion. |
119 |
|
120 |
I don't know cvs at all; hell I hardly know git. If someone is |
121 |
going to carry on with Patrick's work, it's better done sooner, |
122 |
or we may have to do the initial conversion all over again. |
123 |
|
124 |
OFC if you have something to show, by all means push it to the |
125 |
repos: I'd much rather use work you're maintaining than reinvent |
126 |
the wheel. Oh, you'll have to send me an ssh pubkey, as I told you |
127 |
last year, and I'd advise you to keep that off-list. It's nothing |
128 |
to do with the actual work, and simply an infra detail. The |
129 |
ebuild-patch repo[7] is yours, and if you need another for QA |
130 |
or something, just tell me. |
131 |
|
132 |
If again i don't hear from you, then with respect to you as THE |
133 |
uber-user: Do NOT waste any *more* of _my_ time. I don't have a lot |
134 |
of it, and not many years left. And please don't rhapsodise with |
135 |
rationalisations instead of results. I find it intensely annoying, |
136 |
when you didn't just email me to get your repo sorted out months |
137 |
ago. I _expect_ better in future and I *know* you are capable of |
138 |
it. I mean, if you don't take the initiative, then forget about |
139 |
your project, with the best of regards. Who else is supposed to |
140 |
drive it forward, when you clearly don't think it's worth doing? |
141 |
|
142 |
Or y'know: enjoy your private collection on your own, or with |
143 |
others who like working in such a fashion. |
144 |
|
145 |
And good luck to you, truly. |
146 |
|
147 |
Regards, |
148 |
steveL. |
149 |
|
150 |
[1] http://forums.gentoo.org/viewtopic-p-7432716.html#7432716 |
151 |
[2] http://forums.gentoo.org/viewtopic-t-938680.html |
152 |
[3] http://forums.gentoo.org/viewtopic-t-972762.html |
153 |
[4] http://weaver.gentooexperimental.org/trac/gentoo-static/ |
154 |
[5] https://github.com/gentoo/ |
155 |
[6] http://gentooexperimental.org/~patrick/weblog/ |
156 |
archives/2014-02.html#e2014-02-20T09_32_05.txt |
157 |
[7] http://weaver.gentooexperimental.org/trac/ebuild-patch/browser |
158 |
|
159 |
(I hate urls split across email lines: impossible to cp and |
160 |
paste ime. It's current at weblog; paste the relative part |
161 |
if you're reading this in an archive, and care. ;) |
162 |
-- |
163 |
#friendly-coders -- We're friendly, but we're not /that/ friendly ;-) |