[Regression Testing]Questions

[Berillions]

Hi,

I do a regression testing actually for "AC : Brotherhood" because this game crash with wine 1.3.19 and not for old version.

I read the wiki for the regression testing and i have questions for the test.
To begin the test, i launch these command :

Code: Select all

git bisect start
git bisect good wine-1.3.18
git bisect bad wine-1.3.19

For exemple, i have

Code: Select all

Bisecting: 150 revisions left to test after this

After to compile wine and run the game, it crash always. Result, the bug is still present so i launch :

Code: Select all

git bisect bad
Bisecting: 75 revisions left to test after this

I recompile wine and when i test the game, it doesn't crash so i launch :

Code: Select all

git bisect good
Bisecting: 36 revisions left to test after this

My question is :
if the bug is present or not, i must to launch "git bisect bad/good" to see :

Code: Select all

Bisecting: 0 revisions left to test after this

[vitamin]

Bisecting: 0 revisions left to test after this

You have one last compile, test, git bisect good/bad left.

[jjmckenzie]

On 5/8/11 4:08 AM, Berillions wrote:

Hi,

I do a regression testing actually for "AC : Brotherhood" because this game crash with wine 1.3.19 and not for old version.

I read the wiki for the regression testing and i have questions for the test.
To begin the test, i launch these command :

Code:
git bisect start
git bisect good wine-1.3.18
git bisect bad wine-1.3.19



For exemple, i have

Code:
Bisecting: 150 revisions left to test after this



After to compile wine and run the game, it crash always. Result, the bug is still present so i launch :

Code:
git bisect bad
Bisecting: 75 revisions left to test after this



I recompile wine and when i test the game, it doesn't crash so i launch :

Code:
git bisect good
Bisecting: 36 revisions left to test after this



My question is :
if the bug is present or not, i must to launch "git bisect bad/good" to see :

Code:
Bisecting: 0 revisions left to test after this

Keep going as you have to see a 'first bad commit' is: line.

One more time should be enough.

Don't forget to file a bug report if the removing the bad commit causes
the program to run.

James McKenzie

[vitamin]

Hey james, you have the rule now that you ignoring all other posts since you became an admin? You do not need to ask the same question twice.

[DanKegel]

BTW it's a good to verify the endpoints of the
bisect by hand to make sure you're not sending
git off on a wild goose chase.

e.g. don't assume that your distro's wine works the same as
a version you build yourself; instead, build the 'bad' version
you give at the start of git bisect and verify it really is bad.
Likewise for the initial 'good' version.

[jjmckenzie]

Hey james, you have the rule now that you ignoring all other posts since you became an admin? You do not need to ask the same question twice.

I was making the assumption that Berllions had ran the process all the way through. You are correct though that he does have to do one more compile, one more git bisect good/bad and then build again. However, he may have to run through the process again if removal of the identified 'bad commit' does not correct the problem. Dan is also correct in that problems will appear with a pre-build version of Wine. I read more into what you had posted than was supposed to be there. My apologies. And no I did not do this as 'Site Admin'. That is here so that someone other than Dan and Jeremy can remove spammers.

[marceel]

Hi i got this question, I could run game Path of Exile with this version of wine wine-1.5.22-35-g0eea145 but when i downloaded latest wine-git wine-1.5.22-125 and i tried to run it, it crashed. How can i make this regression testing on the same version of wine but different build in git. How can i test it ?

[dimesio]

Use the full version numbers, including the hex digits that identify the most recent commit at the point you are compiling. http://wiki.winehq.org/FAQ#head-dbbb695 ... 08a7713bf3

The first number you give looks like a valid version, but the second one (wine-1.5.22-125) is incomplete.