212940 - Can't use an undefined value as an ARRAY reference at /http/bugzilla/process_bug.cgi line 866.

Status: RESOLVED FIXED

(Bugzilla :: Creating/Changing Bugs, defect)

Description

[Martin J. Evans]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3) Gecko/20030312
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3) Gecko/20030312

Selected multiple bugs to edit, changed the milestone and attempted to accept
the bugs moving to assigned state when this happened:

Content-type: text/html
Software error:

Can't use an undefined value as an ARRAY reference at
/http/bugzilla/process_bug.cgi line 866.

For help, please send mail to the webmaster ([e-mail]), giving this
error message and the time and date of the error. 

Reproducible: Always

Steps to Reproduce:
1. Use the make multiple changes 
2. select multiple NEW bugs
3. accept them and click commit

Actual Results:  
This appeared in the browser:

Content-type: text/html
Software error:

Can't use an undefined value as an ARRAY reference at
/http/bugzilla/process_bug.cgi line 866.

For help, please send mail to the webmaster ([e-mail]), giving this
error message and the time and date of the error. 

Expected Results:  
I expected the multiple NEW bugs I'd selected to be ACCEPTed.

Bugzilla 2.17.4

Comment 1

[Vlad Dascalu]

This might be related/ a duplicate for bug 212962.

Comment 2

[chaduv]

*** Bug 212962 has been marked as a duplicate of this bug. ***

Comment 3

[chaduv]

The relevant lines appear to be:
        if (Param("musthavemilestoneonaccept") &&
                scalar(@{$::target_milestone{$::FORM{'product'}}}) > 1) {

This may be related to the fix for bug 125066, which predates both this and the 
dup. I can't reproduce, but there have been two reports.

Comment 4

[Matisse Enzer]

I get this bug at line 592:

Can't use an undefined value as an ARRAY reference at /path/to/process_bug.cgi
line 592

I am using a patched version of process_bug.cgi
Line 592 is:

     if (Param("musthavemilestoneonaccept") &&

here it is in context:

    /^accept$/ && CheckonComment( "accept" ) && do {
        DoConfirm();
        ChangeStatus('ASSIGNED');
        if (Param("musthavemilestoneonaccept") &&
                scalar(@{$::target_milestone{$::FORM{'product'}}}) > 1) {
            if (Param("usetargetmilestone")) {
                $requiremilestone = 1;
            }
        }
        last SWITCH;
    };

Comment 5

[Matisse Enzer]

Attachment: Crude hack. Defines $::target_milestone{$::FORM{'product'}} if needed

This crude hack simply defines	$::target_milestone{$::FORM{'product'}}   as a
new arrayref if it is not already defeined.

Comment 6

[Ronald Landheer-Cieslak]

Looks like there's another instance of this bug at line 891 of version 2.18rc3
(haven't upgraded to 2.18 proper yet). The attached patch (attachement #1) fixes
the problem for me..

Comment 7

[Jared Warren]

Broken for me to:

Can't use an undefined value as an ARRAY reference at
/old/data/jwarren/web/bugzilla/process_bug.cgi line 879

(was trying to move many bugs from 'New' to 'Accepted' with
"musthavemilestoneonaccept" on

Comment 8

[Jared Warren]

(In reply to comment #7)
> Broken for me to:
> 
> Can't use an undefined value as an ARRAY reference at
> /old/data/jwarren/web/bugzilla/process_bug.cgi line 879
> 
> (was trying to move many bugs from 'New' to 'Accepted' with
> "musthavemilestoneonaccept" on

Sorry, that's on 2.18 proper.

Comment 9

[Max Kanat-Alexander]

All "Software error" messages are critical per justdave.

Comment 10

[Marc Schumann [:Wurblzap]]

Comment on attachment 141064 [details] [diff] [review]
Crude hack. Defines $::target_milestone{$::FORM{'product'}}  if needed

Matisse, assuming you want your patch reviewed, I asked for review for you :)

Comment 11

[Frédéric Buclin]

Comment on attachment 141064 [details] [diff] [review]
Crude hack. Defines $::target_milestone{$::FORM{'product'}}  if needed

IMO, this is the wrong way to fix the problem.

The problem when changing the status of several bugs at once from NEW to
ASSIGNED using musthavemilestoneonaccept=1 is that the product is not defined
and then $::target_milestone{$::FORM{'product'}} is also undefined, which
explains why the system complains.

A better solution is to do this check for each product involved.

Comment 12

[Frédéric Buclin]

Attachment: patch, v1

Check on a per product basis if the target milestone is correctly set. If
several bugs from different products are selected at once and their respective
target milestone is equal to the default milestone of their respective product,
an error message is displayed asking the user to define it first.

Comment 13

[Frédéric Buclin]

Attachment: patch, v2

Updated due to the checkin of a patch from mkanat. :)

Comment 14

[Marc Schumann [:Wurblzap]]

Comment on attachment 174629 [details] [diff] [review]
patch, v2

I think you're addressing it the right way.

>--- process_bug.cgi     17 Feb 2005 21:57:26 -0000      1.235
>+++ process_bug.cgi     17 Feb 2005 23:17:27 -0000
>+        my $nb_milestones =
>+            $dbh->selectrow_array("SELECT COUNT(*) FROM milestones
>+                                   WHERE product_id = ?",
>+                                   undef, $oldhash{'product_id'});

Please make that look at $::target_milestone instead of doing a db lookup and
move r+ forward.

Comment 15

[Frédéric Buclin]

Attachment: patch, v3

Now use
my $nb_milestones = scalar(@{$::target_milestone{$oldhash{'product'}}});
as per Marc's comment.

Comment 16

[Shane H. W. Travis]

2.18:
Checking in process_bug.cgi;
/cvsroot/mozilla/webtools/bugzilla/process_bug.cgi,v  <--  process_bug.cgi
new revision: 1.205.2.13; previous revision: 1.205.2.12
done

Tip:
Checking in process_bug.cgi;
/cvsroot/mozilla/webtools/bugzilla/process_bug.cgi,v  <--  process_bug.cgi
new revision: 1.237; previous revision: 1.236
done

Comment 17

[Max Kanat-Alexander]

The 2.18 tinderbox is burning. This was the checkin that did it.

Comment 18

[Dave Miller [:justdave]]

The error from Tinderbox is:

[Thu Feb 24 07:00:35 2005] process_bug.cgi: Global symbol "$dbh" requires
explicit package name at process_bug.cgi line 1199.
[Thu Feb 24 07:00:35 2005] process_bug.cgi: process_bug.cgi had compilation errors.

Comment 19

[Frédéric Buclin]

Closing this bug again. Marc opened bug 283445 about that. It has a patch ready
for checkin.