Regression testing

[Alexey Loukianov]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

12.04.2012 12:23, Daniel Jeliński wrote:
> ... This time make clean && make depend && make did not help.
> 

I had hit the problems like you describe several times while doing bisecting.
After a lot of trial and error testing I had finally come up with a conclusion
that to get clean and determined bisect results I have to use out-of-tree
build as it is required to delete all the contents of the target build
directory between each and every build. Same stands for WINEPREFIX for the
most part, unless it is known that the problem is not related to the
WINEPREFIX contents (registry, e.t.c.). In case if in doubt it is better to
use fresh prefix each run during bisect process.

So, to summarize, here is the approx. process I use to do bisecting:

1. Determine upper and lower commits bounds to start bisect with.
2. Check out each of them and compare the output of ./configure --help. Check
 if the are any differences between them. If there are differences - check the
commit log for changes to ./configure done between commits in question and
create a separate "invoke configure with a set of options" script. These
scripts would be used later. If there are no significant differences in
configure invocation options - create a single script.
3. Proceed with bisecting, using the following sequence
# WINEBLDDIR=<full path to out-of-tree build directory to use>
# MYWINEPRFX=<full path to the temporarily "testing" wineprefix>
# [ -d $WINEBLDDIR ] && rm -rf $WINEBLDDIR
# [ -d $MYWINEPRFX ] && rm -rf $MYWINEPRFX
# mkdir -p $WINEBLDDIR && cd $WINEBLDDIR
# <execute configure-invocation-script relevant to current configure>
# make -j8
# WINEPREFIX=$MYWINEPRFX ./wine <blablabla>

> I am comparing what I did to [1], and the only difference I see is that I
>  did not run configure before make. I'll try doing that later, but I think
>  this shouldn't affect the outcome. Any other ideas?
> 
> Also I noticed that running parallel make (make -j) is never mentioned on
>  [1]. Is there any reason I shouldn't use it?

I've been using parallel build with WINE for several years up till now and
newer had had any troubles with that. Using CCACHE also seems to be harmful,
so I use it to speed up the bisecting process significantly.

- -- 
Best regards,
Alexey Loukianov                          mailto:mooroon2 at mail.ru
System Engineer,                            Mob.:+7(926)218-1320
*nix Specialist

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.16 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJPhuNrAAoJEPB9BOdTkBULB9MH/3e8sx82qEz/6wjckXOk9XOH
QbbcrDfbkcQnMevfsR9d5TWzHSEDOHOmweFa42vrM0KqD8tk+2YmeHwwbKIHRvrA
ka0hxKuTMVyQ5j7fvdOF19sMpQggEZA5N7fCTA+T2T7AgMp0eMPcMSbCgUDpZbL7
MHF/hwzfIl4SSsLUQkzaOqRXhbhNwKfjx0Sq6v7GVZOg4/eATJFHepMb8/LsqWwt
GE+5Zu8Elk71nDJyxTsehUzloX6+rEPzVKzOh/LEuZF2/t9Y+9rToo5IaQq/fi0f
3eAD8JAZ7ZsPuuKDJXvasLSH3/TY168arHltn25w+RA7ECnI7p3cTvCJ2MJW2z0=
=d4TS
-----END PGP SIGNATURE-----