Avoid infinite loop calling svp_release()

[pmooney]

We hit a bug in our tests when we rolled back twice to the same savepoint. Although this was our bug we only found out when the test never never completed and used much CPU on a Jenkins box.

[ribasushi]

Thank you for the contribution. Is there a chance you could add a failing-without-this test case to the end of https://github.com/dbsrgits/dbix-class/blob/master/t/storage/savepoints.t ?

If you do not have time to do this - that is ok: the issue will be eventually resolved either way, but an accompanying test is a hard prerequisite to applying this change.

[pmooney]

I was hoping you were not going to say that :)

Give me a day or two and I'll add something.

On 10.03.2016 15:46, Peter Rabbitson wrote:

Thank you for the contribution. Is there a chance you could add a
failing-without-this test case to the end of
https://github.com/dbsrgits/dbix-class/blob/master/t/storage/savepoints.t
[1] ?

If you do not have time to do this - that is ok: the issue will be
eventually resolved either way, but an accompanying test is a hard
prerequisite to applying this change.

Reply to this email directly or view it on GitHub [2].

Links:

[1]
https://github.com/dbsrgits/dbix-class/blob/master/t/storage/savepoints.t
[2]
#97 (comment)

[pmooney]

Test added. On the old code it fails, on the new code if passes.

[ribasushi]

This has been applied with a leaner implementation as 7fd61b1fc. The pop is now localised in one spot, and I also removed the alarm-harness, as the test will not be executed on older versions given it is all in one commit.

Thank you for your work!

[ribasushi]

ARGH! I didn't wait for the smokes to finish, failures on older sqlite. Will re-push in a bit, and update the commit SHA

[ribasushi]

Aight, the correct commit is f5f0cb1dd ;)

[ribasushi]

The position of your test (after another block) managed to uncover and lead to a fix of another bug: 66c817df1, so double-cheers ;)

[pmooney]

Glad to be of service :)

The bug in DBIx::Class I fixed highlighted a bug in our own code so it
was actually quite useful ;)

On 30.03.2016 21:37, Peter Rabbitson wrote:

The position of your test (after another block) managed to uncover and
lead to a fix of another bug: 66c817d [1], so double-cheers ;)

You are receiving this because you authored the thread.
Reply to this email directly or view it on GitHub [2]

Links:

[1] 66c817df1
[2]
#97 (comment)

[ribasushi]

The fix is now properly in the CPAN index. Sorry it took so long :/