[Libreoffice-qa] daily bibisect

dk122 at torfree.net dk122 at torfree.net
Mon Jul 29 10:39:10 PDT 2013

Just for convenience, my changes so far today are
They go beyond the changes I mention in-line.

Quoting "Joel Madero" <jmadero.dev at gmail.com>:

> On 07/28/2013 12:00 PM, libreoffice-qa-request at lists.freedesktop.org wrote:
>> I have added some instructions to deal with tags latest and oldest in
>> the daily bibisect repo
>> <https://wiki.documentfoundation.org/index.php?title=QA%2FHowToBibisect&diff=73085&oldid=73084>.
>> Will you be good enough to look over the changes?
>> Of course, if the repo came with these tags laready in place, that
>> whole section of the page could go away.
>> Thanks,
>> Terry.
> I'll take a look - we've been talking quite a bit about cleaning  
> this page and I see that you're adding and cleaning it up. Was  
> wondering if you'd be willing (if not let me know and I'll take a  
> stab at it) to do a couple things to the page.

I will be glad to continue hacking at the page.
> 1. Can you add a note that says if the bibisect shows the bug the  
> entire time (ie. it's prebibisect) to mark the version as 3.5beta0  
> in FDO - this is the best we can do currently - and add  
> "prebibisect" to whiteboard.

> 2. Rearrange the page just a bit - I think logically
> a) Intro
> b) Limitations
> c) Download stuff
> d) how to bibisect (currently in a different location)
> e) Bug that need bibisected
> f) troubleshooting

> I think only one change there but it makes more sense intrinsically to me.
> 3. Another minor thing that I see is in "versions" it says  
> bibisect40bugs for bugs that need bibisecting - as far as I know  
> every bug that needs bibisected is "bibisectrequest" - I haven't  
> seen the 4.0 and 4.0+ (daily) differentiated - perhaps I missed  
> something though :)

BZ shows no bugs with “bibisect40” in either the Whiteboard or
Keywords.  I changed it:

I remember from June that the linked bugzilla queries use a baffling
veriety of selection criteria.  Do you have an idea about whether
there is a reason for this variety?  Or is it something that “just

> Thanks Terry for tackling this how to wiki - it has a lot going on  
> there so the cleaner and simpler we make it the better.

Hah!  We haven't even started on the witheboard statuses and
prescribed comments for reporting the results.  The page already has a
lot of instructions conditioned by the tester's choice of bibisect
download.  I only foresee that getting worse: there are three bibisect
versions, and each can be unhelpful when the regression is off either
end.  (Yes, a regression newer than the 40+ version is plausible.  I
went weeks without updating my download after an attempted update hung
for days showing an ETA of a hundred odd days.)  Among the unseccssful
cases, only the determination that a regression is older than the OVA
or 40 repositories marks the end of the search.  How much should we
try to track in the whiteboard status?

> If you're busy or just don't want to do it just let me know - I  
> didn't want to be editing over you :)

I am happy to continue.  Knowing that someone is checking my work
frees me to make bigger changes.


More information about the Libreoffice-qa mailing list