1 |
>>>>> On Mon, 03 Dec 2018, grozin wrote: |
2 |
|
3 |
> On Sun, 2 Dec 2018, Michał Górny wrote: |
4 |
|
5 |
>> I think that if there's one package that doesn't work with profiles |
6 |
>> (compared to the very large number of packages which just work fine), |
7 |
>> it's not the profiles but the package being broken (read: doing silly |
8 |
>> assumptions). Therefore, it's not 17.0 profiles being the problem |
9 |
>> but the package in question. |
10 |
>> |
11 |
>> Claiming that people doing any change to Gentoo are required to fix |
12 |
>> all the problematic packages is just silly. This is basically saying |
13 |
>> that it's fine to add bad quality packages and then demand others to |
14 |
>> fix them for you. People who worked on the profile can fix bugs in |
15 |
>> the profile. Don't expect them to pursue whatever broken packages |
16 |
>> you like just because they happened to change the fragile conditions |
17 |
>> under which they worked. |
18 |
|
19 |
Oh, come on. The 17.0 profiles introduced rather daring compiler and |
20 |
linker options, and clozurecl is not the only package broken by them. |
21 |
Most of the Lisp packages (including Emacs) are affected, because their |
22 |
dumping of the executable is incompatible with PIE. That doesn't make |
23 |
them "bad quality packages". It simply means that the PIE flags hadn't |
24 |
previously been encountered upstream, or not reported to them. |
25 |
|
26 |
> See bug #672454. |
27 |
|
28 |
> clozurecl compiles and works fine with the upstream-provided |
29 |
> compilation flags. So, we cannot ask the upstream to solve our |
30 |
> problems for us. |
31 |
|
32 |
Still, you could report it upstream, maybe with a patch for their build |
33 |
system? |
34 |
|
35 |
Ulrich |