#577729 debbugs: The 'bug' parameter ... did not pass regex check

#577729#5
Date:
2010-04-13 23:24:41 UTC
From:
To:
I think these error messages can be improved:

| Processing commands for control@bugs.debian.org:
|
| > clone 577650 -1
| Bug#577650: libcupt-perl: Does not handle automatic upgrade of git packages
| Bug is marked as being merged with others. Use an existing clone.

Huh?  Does that mean I have to unmerge before cloning?  Once I understand
this, I’ll be happy to file a separate bug for it.

| > retitle -1 conflict on git-completion should be versioned to avoid dpkg bug
| Failed to set the title of -1: The 'bug' parameter ("-1") to Debbugs::Control::set_title did not pass regex check

I would prefer something to the effect of

  Failed to set the title of -1: undefined psuedo-bug-number -1

or

  Failed to set the title of -1: no such bug, since corresponding clone failed

| 	Debbugs::Control::set_title('transcript', 'IO::Scalar=GLOB(0x31d4ce8)', 'requester', 'Jonathan Nieder <jrnieder@gmail.com>', 'request_addr', 'control@bugs.debian.org', 'request_msgid', '<20100413224129.GA1922@progeny.tock>', 'request_subject', ...) called at /usr/lib/debbugs/service line 879
| 	eval {...} called at /usr/lib/debbugs/service line 878

Probably the debugging output should be suppressed, too.

[...]
| > thanks
| Stopping processing here.

Thoughts?
Jonathan

#577729#10
Date:
2010-04-13 23:39:06 UTC
From:
To:
Jonathan Nieder wrote:

Ah, found it:

http://bugs.debian.org/474742

This is super counter-intuitive.  So let me add another item to this
error message wishlist (that makes 3):

I think this should say:

  Bug is marked as being merged with others. Please unmerge and reassign
  or unmerge, clone, and merge again.

Why would I be asking to clone if there were an existing clone to use?

HTH,
Jonathan

#577729#15
Date:
2010-04-13 23:58:34 UTC
From:
To:
forcemerge 577487 57729
thanks

If the bug is merged, unmerge it, and use one of the existing bugs.

[...]


The debugging output is there so that I know what is going on when
people report things, so no, it won't be suppressed.


Don Armstrong

#577729#20
Date:
2010-04-14 05:17:07 UTC
From:
To:
Don Armstrong wrote:

Maybe the message can make that more clear.  For example, what did you
think of this suggestion?

 -Bug is marked as being merged with others. Use an existing clone.
 +Bug is marked as being merged with others. Please unmerge and reassign
 +or unmerge, clone, and merge again.

Yes, it is correct, but “did not pass regex check” is really not
helpful.

How about “-1 isn't a valid bug number”?

Of course, I only meant suppressing the debug output in this well
understood case.  It was only a wishlist bug, though, so if you like
the debugging output, that is fine.

BTW, I would be glad to prepare a patch if directed to the right
source tree to start from.

HTH,
Jonathan

#577729#25
Date:
2010-04-14 06:45:30 UTC
From:
To:
I'd prefer not to add more verbiage there, and stick it in the
documentation instead. [And it certainly shouldn't say to reassign;
the critical aspect here is just that you just unmerge and use it...
thus, use an existing clone.]

Because the test is done using a regex, and any failure in the options
passed to that control implementation would then fail with that
message, which would be incorrect.

http://bugs.debian.org/debbugs-source/debbugs and mainline.


Don Armstrong

#577729#30
Date:
2010-04-14 07:52:15 UTC
From:
To:
Don Armstrong wrote:

Thanks.  I assume this should be

http://bugs.debian.org/debbugs-source/debian

and that I should fetch with bzr.

Also thanks for the other explanations.  I’ll look at this some
more tomorrow.

Regards,
Jonathan