1 |
On Tue, 6 Aug 2013 17:30:56 +0100 |
2 |
Ciaran McCreesh <ciaran.mccreesh@××××××××××.com> wrote: |
3 |
|
4 |
> On Tue, 6 Aug 2013 12:16:57 -0400 |
5 |
> Alexis Ballier <aballier@g.o> wrote: |
6 |
> > On Tue, 6 Aug 2013 16:25:12 +0100 |
7 |
> > Ciaran McCreesh <ciaran.mccreesh@××××××××××.com> wrote: |
8 |
> > > On Mon, 5 Aug 2013 18:49:49 -0400 |
9 |
> > > Alexis Ballier <aballier@g.o> wrote: |
10 |
> > > > Again, symbols have nothing to do here. Since you have poor |
11 |
> > > > control on overlays and absolutely zero control on locally built |
12 |
> > > > packages, the only sane assumption is that all symbols are used. |
13 |
> > > |
14 |
> > > And that answers your question as to why the spec says "may". |
15 |
> > |
16 |
> > Not really: This explains why there _must_ be another way than |
17 |
> > subslots to check this. |
18 |
> |
19 |
> Only if you want to avoid ever having an unnecessary rebuild. That |
20 |
> isn't a sensible goal. |
21 |
|
22 |
|
23 |
You haven't read the part you cut it seems. |
24 |
|
25 |
|
26 |
> > > But that leads to breakage when the "stable" ABI changes. It's |
27 |
> > > better to avoid breakage than it is to require the odd extra |
28 |
> > > recompile. |
29 |
> > |
30 |
> > That's why preserve-libs is still the proper way to do it until we |
31 |
> > have a more accurate subslot handling. |
32 |
> |
33 |
> preserve-libs is broken by design. All you need is proper use of |
34 |
> subslots, and to recognise that the occasional unnecessary rebuild |
35 |
> isn't a big deal. |
36 |
|
37 |
preserve-libs has its flaws. All you need is improving the subslot |
38 |
system so that it can be used properly. What are you waiting for ? :) |
39 |
|
40 |
'occasional unnecessary rebuild' is a big deal since subslots introduce |
41 |
this regression... |
42 |
|
43 |
> > In the poppler case, the |
44 |
> > stable ABI changes an order of magnitude less often than the |
45 |
> > unstable one; it's not really 'the' extra recompile but rather |
46 |
> > hundreds of them if you multiply by the # of dependent packages. |
47 |
> > Claiming it's better to have hundreds of recompiles is simply being |
48 |
> > lazy, throwing our failures onto users, and not be willing to fix |
49 |
> > the real problem. |
50 |
> |
51 |
> There's an easy fix for that: split the package up. |
52 |
|
53 |
|
54 |
Great, please start submitting patches at our thousands of upstreams so |
55 |
that their packages can be split properly. |
56 |
|
57 |
|
58 |
> > > With subslots, the developer specifies dependencies. With |
59 |
> > > fix-linkage, the package mangler has to guess based upon very |
60 |
> > > incomplete information. |
61 |
> > |
62 |
> > You're assuming library maintainers are stupid: If done properly, |
63 |
> > there is a library that gives you the text file location based on |
64 |
> > where it has been installed. If that's not the case then that's |
65 |
> > where I'd try to improve things, not hiding the problem behind a |
66 |
> > subslot. |
67 |
> |
68 |
> That's no good when that file is removed or overwritten by a newer |
69 |
> format, but preserve-libs leaves the .so that uses that file behind. |
70 |
|
71 |
Then submit patches to run preserve-libs as soon as possible as I |
72 |
suggested in the part you cut. preserve-libs, in contrast to just |
73 |
removing the .so, leaves you with a working system in 90% of the cases. |
74 |
|
75 |
Exercise: Try to update a FreeBSD 6 system (libc.so.6) to a FreeBSD 7 |
76 |
system (libc.so.7) without some kind of preserve-libs mechanism. |
77 |
Been there, done that. The flaws of preserve-libs show up but it |
78 |
maintains a half working system all the way long that allows you to |
79 |
finish the update. |