[Libreoffice-qa] Office XP/2003, 2007/2010 formats import/export -- most used feature.

[dE .]

On 04/23/12 20:47, Michael Meeks wrote:
> Hi there,
>
> On Mon, 2012-04-23 at 19:04 +0530, dE . wrote:
>> The MAB has been chaotic lately... actually it turned out that I didn't
>> get a lot of mails and missed out a lot of comments following the LONG
>> discussion in the bug... so sorry about that, I noticed them now.
> 	:-)
>
>> What I've been trying to add to MAB are broken import/export
>> features/regressions of Office XP/2003, 2007/2010 formats.
> 	While I disagree with your analysis of the use of ODF, I agree that
> fixing interoperability bugs is important. Having said that, I'd really
> like to keep the number of most annoying bugs really low.

When I say the usage is very low, I'm looking at the statistics of 
google search results and the marketing tactics of commercial office suits.

Looking at these I find use of ODF only in opensource communities. 
Practically it has not been adopted anywhere, and that's why commercial 
office suits (who are more business oriented) do not give much 
preference to ODF.

>> These bugs are more critical than a few crashes and I estimate there're
>> ~200 bugs in LO's bugzilla alone.
> 	If there are more than about 20 open MAB (currently there are ~70) it
> becomes extremely easy to miss the screamingly urgent bugs that this was
> setup to track. If you add another 200 there, no-one wins and we just
> destroy this valuable place to look for "real blockers" :-)
>
> 	There is some sort of near-zero-sum-game here, and hiding the signal in
> a ton of noise is not a good idea. Worse - without developer input, what
> looks to you like a simple "inter-operability bug" may in fact be a
> substantial feature that needs a man month to implement, so serious care
> is needed with these.
>
> 	Having said that, having a keyword for inter-operability problems, that
> we track the queries of, and try to increase interest in could be rather
> good. IIRC we had an interop whiteboard status at some stage ? and we
> could track that.
>
> 	As a final thought - playing with the 'Most Annoying Bugs' is really
> the pinacle of the QA effort :-) I'd suggest that for each person adding
> or removing a bug from there, they also should do a handful of the more
> vanilla work: checking and moving from UNCONFIRMED to a suitable state
> eg. :-)
>
> 	Just my 2 cents,
>
> 		Michael.
>

Yes, absolutely, the MAB should be in a reasonable limit such that it 
looks 'solvable'.

What I suggest is a metabug for interoperability issues. These issues 
are hard to fix, and the dev team should know that these bugs, although 
hard to fix, are very important and if some dev does have the skills to 
fix them, he should fix these bugs instead of doing something else.

Also in the list of ~200 bugs, I think real bugs will be only in ~50.. 
the others must be duplicates; the same bug reproducing itself in many 
way. But these duplicates can only be detected by developers themselves 
since this requires knowledge of the format.