1 |
On Mon, 30 Sep 2013 08:14:29 +0200 |
2 |
Agostino Sarubbo <ago@g.o> wrote: |
3 |
|
4 |
> These type of failures are _not_ architecture dependant. |
5 |
|
6 |
This is wrong. |
7 |
|
8 |
Libraries behave differently on different architectures because the |
9 |
compiled code is actually different. Different architectures use |
10 |
different ways to access and manipulate memory. Libraries are |
11 |
linked/loaded differently on different architectures. |
12 |
|
13 |
If there is architecture specific code in a library, there is a really |
14 |
good chance that the internal test suite does not cover it because |
15 |
upstream doesn't have the hardware to test it. |
16 |
|
17 |
If there are generic code paths in a library that gets used when no |
18 |
architecture-specific code is available for a specific architecture, |
19 |
those might go untested upstream as well as on the package maintainer's |
20 |
system, since some (optional?) alternative architecture specific |
21 |
optimisation code paths might be (automatically) selected in either |
22 |
case. |
23 |
|
24 |
I could go on... |
25 |
|
26 |
> So, instead of have 10 ATs that are testing the same rdeps, seems |
27 |
> logic that the maintainer could do it one time. |
28 |
|
29 |
Your logic is flawed. |
30 |
|
31 |
> What do you think? |
32 |
|
33 |
I think that when you set out to help every minor architecture get |
34 |
stable, you didn't know what you were getting into. |
35 |
|
36 |
|
37 |
jer |