1 |
On 7/10/06, Ryan Hill <dirtyepic.sk@×××××.com> wrote: |
2 |
|
3 |
> Ebuilds shouldn't die on anything according to the non-interactive portage |
4 |
> philosophy. I don't know how official that philosophy is though. |
5 |
|
6 |
Correct me if I'm wrong, but this has nothing to do with being |
7 |
interactive or not. To me, an ebuild that dies (intentionally or due |
8 |
to a build error) isn't interactive at all. |
9 |
|
10 |
> > 1) Should all ebuilds that currently filter --fast-math die on its |
11 |
> > presence instead of filtering it ? |
12 |
> |
13 |
> No, that would be a major pain in the ass for anyone wanting to use -fast-math, |
14 |
> which does have legitimate uses. |
15 |
|
16 |
If a package is known not to work with a certain flag, being able to |
17 |
emerge it won't change the fact that it doesn't run. Plus, if the |
18 |
solution is considered good for python, I don't understand why it |
19 |
wouldn't be good for any other package. Are you saying that Paul's |
20 |
proposition of having the python ebuild die on use of --fast-math |
21 |
isn't good ? If yes, why ? And what is your better idea ? |
22 |
|
23 |
> > 2) If yes, are there any other flags that ebuilds should die on ? |
24 |
> |
25 |
> There's a million, and they're constantly changing. For example, |
26 |
> -frename-registers is generally safe on GCC 3.4, broken in 4.0, and enabled by |
27 |
> default on 4.1. |
28 |
|
29 |
Which is exactly the reason why we could benefit of something that |
30 |
enables us to manage this in a clean and safe way. I'm not saying I |
31 |
have a candidate for that "something", but I wanted to discuss if |
32 |
there was an interest in it. |
33 |
|
34 |
Let's take again the example of -ftracer which can be enabled by the |
35 |
-fprofile-use meta-flag. Imagine we have a mechanism somewhere (again, |
36 |
the reason I'm being vague is that my point isn't to discuss |
37 |
implementation just yet) that adds -fno-tracer to CFLAGS. In this |
38 |
case, you're covered wether -ftracer was added directly on indirectly |
39 |
by fprofile-use, which actually simplifies the number of flags that |
40 |
you need to blacklist. Thus ebuilds don't have to take care of it, |
41 |
bugs don't pour into bugzie, and Jakub can avoid overheating. |
42 |
|
43 |
> Users playing with CFLAGS get to keep the pieces. Trying to dummy-proof the |
44 |
> system doesn't help anyone but the dummies. ;) |
45 |
|
46 |
I'm one of those devs who care for our users. I think it's dangerous |
47 |
to try and categorize users in, for example, dummies and non-dummies, |
48 |
as you say. Who are we to judge this or that user is a dummy ? Plus, |
49 |
we all are the dummy of somebody else. |
50 |
|
51 |
Anyway, I was thinking more in terms of making the job of developers, |
52 |
bug wranglers, and poor old bugzilla easier, cleaner, safer. How many |
53 |
bugs do we have that are due to dangerous flags ? How much time and |
54 |
effort could we save if we didn't have those ? Also, I was thinking |
55 |
that if a good solution was found to deal with a dangerous flag in a |
56 |
certain package, maybe it was a good idea to extend this solution to |
57 |
other packages. And finally, if said solution becomes common, maybe |
58 |
it's a good idea to make it system-wide with a possibility to override |
59 |
the setting by the user or the ebuild. It seems we already have |
60 |
per-package CFLAGS, so part of this, at least, is already implemented. |
61 |
|
62 |
|
63 |
On 7/10/06, Richard Fish <bigfish@××××××××××.org> wrote: |
64 |
|
65 |
> My (user) opinion is that ebuilds should not die on CFLAGS, at least |
66 |
> not until per-package CFLAGS are implemented. |
67 |
|
68 |
Why ? Stating your opinion without any justification isn't really |
69 |
constructive. And same as above : being able to emerge a package that |
70 |
won't run doesn't help you more. |
71 |
|
72 |
> Now if someone is crazy enough to enable -ffast-math globally or |
73 |
> specifically for python in that situation, well, die, spin the cpu fan |
74 |
> backwards, melt the hard disk down and sell it for scrap, whatever you |
75 |
> want! |
76 |
|
77 |
Same as above again, replace 'dummy' with 'crazy'. |
78 |
|
79 |
Denis. |
80 |
-- |
81 |
gentoo-dev@g.o mailing list |