Re: Closing resolved bugs

[Cyrille Chepelov <chepelov calixo net>]

Le ven, sep 07, 2001, à 08:39:53 -0700, Robert Campbell a écrit:

> There are quite a few resolved but unclosed bugs on Bugzilla.  Am I right 
> in assuming that they only need to be verified by a 3rd party before they 
> can be closed?  I would be happy to check out as many as I can, if this 
> is true.  

My plan was to mass-close all "resolved" bugs which apply to < 0.88.x 
when 0.89 is out, and to do some checking for 0.88.x bugs. Now, I'm very
happy someone volunteers, and since I gather from your message that you
already have bugzilla write access, don't hesitate.

FWIW, there are some bugs I mark outright INVALID:
	* bugs not in English
	* bugs without a valid return address. I don't read what Anonymous
Cowards say.
	* bugs for versions 0.86 or older.
(always with a short, perhaps not very diplomatic, but direct explanation).

these probably don't need much effort to be verified.

I plan to consider 0.88.x obsolete two or three months after the release of
0.89 (depending on the number of crasher bugs we find in 0.89 release, and
perhaps the number of releases beyond 0.89 in that 2-3 month window), and
mark as invalid crashbugs against 0.88.x. 

Here are the two most common duplicates:

#50750: anti-aliasing crashes (fixed ?)
#52836: tools..foo crashes in context menu (fixed after 0.88.1)
        (dupes: 54976 55428 56753 57014 58329)

(nowadays, #50750 has almost completely faded out. #52836 is alive and well)

Also, I was about to mark as "RESOLVED REPORTERTHINKSDEVELOPERISHISSLAVEORDOG" 
all bugs marked as NEEDINFO for more than, say, 4 weeks without answer to 
a question.



> Some guidelines would be helpful.  For instance, I am running the Windows 
> version on NT 4.0.  Can I close bugs that were reported on Linux that I 
> have verified fixed on Windows?  The other project that I follow closely is 
> AbiWord.  Parts of its code can be very different on different platforms, 
> so I do not close a bug reported on another platform unless it has been 
> reproduced on Windows.  But Windows and *NIX versions of Dia are nearly 
> identical, correct?  So would it be OK to close bugs reported on Linux 
> (for instance) that I have verified fixed on Windows?  Some bugs clearly 
> wouldn't apply to Windows, so I would of course be reasonable.

Two areas need closer attention: 
	* whatever touches the file system (by definition)
	* whenever there is a smell of I18N (the GTK used by the Win32 port
talks UTF-8, the stable *nix gtk speaks the local 8-bit charset. Dia talks
local8 too, except some more or less well insulated modules which already
talk UTF-8 (in 0.88.x, the PostScript Unicoder and that's about it, in 0.89
add XML handling, the construction of the stereotype string ("<< foo >>"),
and some bits of the StdProp code). Trouble on Win32 is that dia talks
local8 to a component which talks UTF-8, and there are not many steps done
to resolve the impedance mismatch (well, I'm into a long-term audit, to make
that a thing of the past. Please check #55904 for some details...).
 	And, of course, I don't know how xml-i18n-tools fare under Win32
(Hans, a comment ?)

And of course, we don't have a WMF export plug-in for !Win32 (but you're on
the right side of the fence wrt this problem). This might become an area of
mismatch if someone uses one of the libraries now available to write a WMF
export plug-in for Linux.	

Good luck, and many thanks for your help !

	-- Cyrille

-- 
Grumpf.