1 |
On Sun, 2 Jul 2017 17:22:03 +0200 |
2 |
Toralf Förster <toralf@g.o> wrote: |
3 |
|
4 |
> A lot of findings did appear immediately. |
5 |
|
6 |
Your stats roughly coincide with my own, but I test only the |
7 |
intersection of changes that mention "perl" somewhere ( primarily as an |
8 |
effort to make it so I can get perl stuff in good enough condition that |
9 |
we can use the assumption of passing so we can to full-tree regression |
10 |
tests against development releases of Perl, to minimise the amount of |
11 |
release-panic we suffer, and to get data to make upstream more aware of |
12 |
what is getting broken in advance of it happening ) |
13 |
|
14 |
|
15 |
3915/4252 -> 92.07% ( 337 todo, 360 broken (8.47%), 92 to report) |
16 |
|
17 |
Also, tip: |
18 |
|
19 |
Instead of doing |
20 |
|
21 |
LISTGENERATOR | xargs |
22 |
|
23 |
Do: |
24 |
|
25 |
LISTGENERATOR > file ; xargs -a file |
26 |
|
27 |
Benefits: |
28 |
|
29 |
- You can look at your todo list manually if you have to |
30 |
- STDIN doesn't get closed/redirected for portage |
31 |
|
32 |
The latter is pretty handy, as there's occasionally packages that |
33 |
behave differently if they're attached somewhere to a real human input |
34 |
source. |
35 |
|
36 |
Some don't prompt for input, but make assumptions about who is |
37 |
watching based on the ability to read from STDIN ( specifically, asking |
38 |
if STDIN is a TTY or a PIPE ). |
39 |
|
40 |
Others may have devious code that expects interactive behaviour, with a |
41 |
fallback that only takes place if STDIN is closed/not a TTY. |
42 |
|
43 |
TL;DR: Exposes different kinds of fun bugs. |