← Back to team overview

launchpad-dev team mailing list archive

Re: bug message indices. stable allocation and hidden messages

 

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 1/25/2011 4:13 PM, Robert Collins wrote:
> So, one of our deeper problems is that BugTask:+index has to load
> -every- bug message to be able to show any bug message, because we
> need to know the index of the message in the page.
> 
> There are a few subtleties around this and I'd like it if folk with
> either strong UI opinions or knowledge of gotchas in the bugs code
> could comment on the proposed changes below.
> 
> To fix this we need to denormalise the index and instead of
> calculating it on the fly, use it as something we can query on.
> 
> Beyond that we have a choice:
>  - we can keep the current rules for message index - that is that
> hidden messages lose their index, and that bug syncing can change the
> timeline and make messages appear in the past
>  - we can change the rules, so that message indices become stable -
> once allocated never changed.
> 
> I've chatted with Deryck in the past and we concluded that the second
> option is ok UI wise (indeed, its preferrable). Code and performance
> wise the second option is a clear winner because we don't have to
> rewrite indices that have been allocated. We *may* need to make some
> UI changes as a result of not having a consecutive series - but I
> don't think so because we already support showing first + last N,
> which is effectively the same.
> 
> Does anyone see an issue with us making bug message index allocation write-once?

Why can't it be written as part of the object, and if you want to insert
a message into the middle, you have to rewrite all of the rest? I guess
I don't really see why it has to be write-once.


> 
> Separately, there is an open question about how to allocate the
> indices when there are already hidden messages; I propose to be very
> simple and just do date first->last.
> 
> If noone comments I should have a patch up allocating this stuff tomorrow.
> 
> -Rob

So is the issue that if there are hidden messages you shouldn't get gaps
in the numbering?

Is it something that some users can see these hidden messages and some
can't? What is the handle on a "hidden" message for people who can
actually see it?
John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk0/UrYACgkQJdeBCYSNAAO5OwCg0doo6NM0LnGwx/YF+YzahETY
UIcAoNgJZ43XfU3A9scKwoMVB9dTCGBy
=c73o
-----END PGP SIGNATURE-----



Follow ups

References