Last Comment Bug 480968 - Error message popup on Windows from Oracle module on every run
: Error message popup on Windows from Oracle module on every run
Product: Bugzilla
Classification: Server Software
Component: Installation & Upgrading (show other bugs)
: 3.3.3
: All Windows XP
: P1 major with 1 vote (vote)
: Bugzilla 3.4
Assigned To: Max Kanat-Alexander
: default-qa
Depends on:
  Show dependency treegraph
Reported: 2009-03-02 08:51 PST by Marc Schumann [:Wurblzap]
Modified: 2010-02-28 10:44 PST (History)
3 users (show)
mkanat: approval+
mkanat: blocking3.6+
mkanat: approval3.4+
LpSolit: blocking3.4-
See Also:
QA Whiteboard:
Iteration: ---
Points: ---

v1 (1.76 KB, patch)
2010-01-07 14:02 PST, Max Kanat-Alexander
wurblzap: review+
Details | Diff | Splinter Review
v1 (3.4) (2.77 KB, patch)
2010-01-08 09:35 PST, Max Kanat-Alexander
wurblzap: review+
Details | Diff | Splinter Review

Description Marc Schumann [:Wurblzap] 2009-03-02 08:51:06 PST
Every run yields a pop-up on Windows with ActiveState Perl 5.10.0, saying "This application has failed to start because OCI.dll was not found. Re-installing the application may fix this problem." Workaround is to simply click Ok.

This happens even without any intention to install Oracle -- I never installed DBD-Oracle, I think it came in with the ActiveState Perl install (ppm tells me the module is in "perl", not in "site").

I'm aware this can probably be cured by deinstalling DBD-Oracle. I still think we should do something about it because it'll probably bite *lots* of people after our next release.
Comment 1 Frédéric Buclin 2009-03-02 09:32:51 PST
IMO, this is not a Bugzilla bug. I fixed the problem by typing (from memory):

ppm --remove --area perl DBD::Oracle

This problem should be reported to ActiveState, who installs it by default with no reason. IMO, this bug is INVALID (not a Bugzilla bug).
Comment 2 Max Kanat-Alexander 2009-03-02 11:11:32 PST
It would require extensive amounts of workaround code just to fix this, which is (I agree with LpSolit) a bug in the package that ActiveState ships for DBD::Oracle.

FWIW, it already affects people, our next release won't make any difference.
Comment 3 Max Kanat-Alexander 2009-03-02 11:22:32 PST
I've reported this bug to ActiveState:
Comment 4 Teemu Mannermaa (:wicked) 2009-04-14 11:51:41 PDT
Shouldn't we implement either of the workarounds suggested in the ActiveState bug report? Namely either 1) use to determine the version number of installed modules, or 2) call SetErrorMode() to disable Windows error dialogs.
There's even working code for the latter case in the ActiveState bug report. And former seems something should probably always do if that's somehow "better" method to determine versions or fixes some other symptoms of module loading we might have had to workaround in the past.
Or is there a known reason neither of those can be used in Bugzilla?
Comment 5 Max Kanat-Alexander 2009-04-14 15:14:28 PDT
Yeah, we could do the SetErrorMode thing.
Comment 6 Max Kanat-Alexander 2010-01-07 13:53:49 PST
BTW, the reason that we don't want to use to determine versions is that we want to do *exactly* what Bugzilla is going to do when loading the module, so that we know that it will in fact load and work and that all its prerequisites are installed--not just that it itself is on the filesystem somewhere.
Comment 7 Max Kanat-Alexander 2010-01-07 14:02:41 PST
Created attachment 420620 [details] [diff] [review]

This is the workaround recommended by ActiveState.
Comment 8 Max Kanat-Alexander 2010-01-07 14:19:31 PST
Comment on attachment 420620 [details] [diff] [review]

I can't test this, since apparently my installation doesn't have DBD-Oracle and I can't install it via PPM, but I do know that this compiles, and it's basically identical to the code from ActiveState.
Comment 9 Frédéric Buclin 2010-01-07 15:13:57 PST
(In reply to comment #8)
> (From update of attachment 420620 [details] [diff] [review])
> I can't test this, since apparently my installation doesn't have DBD-Oracle and
> I can't install it via PPM

I have the same problem. Once it's removed, it seems impossible to reinstall it.
Comment 10 Max Kanat-Alexander 2010-01-07 15:30:32 PST
Comment on attachment 420620 [details] [diff] [review]

Hmmm. Byron, do you want to help us out with our dilemma here? :-)
Comment 11 Marc Schumann [:Wurblzap] 2010-01-08 02:11:38 PST
Comment on attachment 420620 [details] [diff] [review]

This resolves the problem for me.
Comment 12 Max Kanat-Alexander 2010-01-08 09:29:46 PST
Comment on attachment 420620 [details] [diff] [review]

Awesome; that's good enough for me.
Comment 13 Max Kanat-Alexander 2010-01-08 09:35:44 PST
Created attachment 420748 [details] [diff] [review]
v1 (3.4)

3.4 requires a different patch because it doesn't have init_console().
Comment 14 Marc Schumann [:Wurblzap] 2010-01-08 13:15:21 PST
Comment on attachment 420748 [details] [diff] [review]
v1 (3.4)

Yup, this works for me.
Comment 15 Max Kanat-Alexander 2010-01-08 13:24:54 PST
Awesome sauce.
Comment 16 Max Kanat-Alexander 2010-01-31 09:39:59 PST

Checking in Bugzilla/Install/;
/cvsroot/mozilla/webtools/bugzilla/Bugzilla/Install/,v  <--
new revision: 1.26; previous revision: 1.25


Checking in;
/cvsroot/mozilla/webtools/bugzilla/,v  <--
new revision: 1.561.2.1; previous revision: 1.561
Checking in Bugzilla/Install/;
/cvsroot/mozilla/webtools/bugzilla/Bugzilla/Install/,v  <--
new revision:; previous revision:

Note You need to log in before you can comment on or make changes to this bug.