feature requests for loggerhead

[John Arbash Meinel]

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

Michael Hudson wrote:
> Aaron Bentley wrote:
>> Michael Hudson wrote:
>>
>>> Fixing
>>> performance and reliability issues are the main things I know about, but
>>> I'd like to here from people who use or would like to use
>>> loggerhead/codebrowse and wish it did things differently.
>> It would be nice if it showed the exact changes, rather than
>> highlighting the entire line.  This can be done by comparing
>> non-matching blocks.
> 
> That would be nice, yes.  difflib has code for this, I think, does
> bzrlib have it somewhere too already?

I believe you can use the same infrastructure, but you have to do a bit of it
"by hand".

If you know that 2 lines "mostly match" then you can do a diff on the strings
(rather than a list of lines). And it will tell you what characters match/don't
match.

The PatienceSequenceMatcher uses the same api as difflib.SequenceMatcher (in
fact it inherits from it, and only implements the core matching function).



> 
>> It would also be nice if all code views had
>> annotation information, since it is so cheap to retrieve.
> 
> You mean in the diff views too?  I'd have thought the concern there
> would be space to put it in, rather than cost of fetching it.

Well, you could annotate using dotted revnos (and maxed at a certain number of
characters), where they can be links/have tooltips for more information.

> 
>> It would be nice to provide "bzr viz"-style output.  I understand the
>> mercurial guys ported bzr viz to their web interface.
> 
> Using dot or similar?  That occurred to me to.
> 
> Browsing around the one mecurial repo I know the URL of didn't turn up
> anything like what you describe though, can you provide a link?

I think there is also a way in Cairo to render to a file rather than to screen.
So you can re-use the existing code, rather than having to generate dot files.

The problem with dot (that *I* have) is that it generates really huge nodes. So
it is hard to have a lot of them on a single page.

> 
>> In fact, the loggerhead output doesn't provide an easy way to see what
>> revisions were merged, and I think it would be great if it did.
> 
> It's in the stuff that drops down when you click the arrow between the
> revno and the commit message, or are you talking about something else?
> 
>> It would be cute to include hackergochi in the log output, including all
>> relevant committers in merge revisions.  PQM could be a robot, e.g.
>> Robby the robot.
> 
> Too cute for now, I think :-)
> 

Well, we already have Hackergotchies for Launchpad, it would just be a matter
of *using* them in codebrowse.

>> Scanning the text of changes for TODOS, FIXMEs and NEWS entries could
>> also provide interesting results.
> 
> Yes, but for now that seems a bit DWIMish for me.
> 
> Cheers,
> mwh
> 
> 

Well, if you look at
https://code.launchpad.net/~jameinel/bzr/log-short-merges-83887

You can see that Launchpad is already searching through the commit messages,
and generating hyperlinks for stuff like "bug #83887".

I understand you "DWIM" issues, but "(?i)(bug #|bug |#)(\d+)" seems reasonable.
Though we would have to watch out for regular comments so that:

#I'm mentioning that I have
#10000 different widgets

John
=:->

PS> I realize this is separate, but it would be really nice if going to:
https://code.launchpad.net/~jameinel/bzr

would show me all of my branches of bzr only. It follows naturally since:
https://code.launchpad.net/~jameinel/bzr/BRANCH is the branch location. and
https://code.launchpad.net/~jameinel/ is the list of all of my branches.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGdrQ7JdeBCYSNAAMRApJ+AKCefvP5YO7ufYPpHHvFhXKqQRXTcACeMIe3
i/5TRuTrJPITYo3TFevfWhg=
=vNZ0
-----END PGP SIGNATURE-----