feature requests for loggerhead
Kent Gibson
warthog618 at gmail.com
Tue Jun 19 14:44:06 BST 2007
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Martin Pool wrote:
>
>
> * The display should be more dense - loggerhead uses (i think) a
> larger-than-default font, and has a lot of whitespace. (I like
> whitespace but somehow in the current theme it looks gassy rather than
> elegant.) For example in my browser window I see 19 revisions by
> default of gtk+ at the url Johan gave, but only 5 of bzr.dev, and I
> don't think the viewvc is too dense.
>
I also got tired of the gassy changes view so I had a hack at
loggerhead myself a few months back to produce an output more along
the lines of bzr log --line.
Here are a couple of screenshots -
http://members.iinet.net.au/~warthog/ss0.png and
http://members.iinet.net.au/~warthog/ss1.png
(sorry about that - I don't have a live site available and can't for
the life of me workout how to capture the pages in HTML including the
style sheet).
The triangle expands the revision out to something more along the
lines of bzr log --long, as shown in the second screenshot.
I dropped the committer from the collapsed view cos I don't really
care who checked it in (for bzr.dev it is usually PQM anyway), and if
you are really interested you can find it in the expanded view.
As you can see I get 25 revisions to the page (collapsed view). In
the same window size I only get 8 from vanilla loggerhead.
The version I setup at work also has a 'home' button on the toolbar to
take you back to the branches front page.
Kid was also mentioned in the thread. That is definitely a primary
cause of slowness in loggerhead. In the benchmarking I've done
(trying to track down why generating the revision view was so damn
slow) I found most time was spent in "turbokid.kidsupport: Applying
template loggerhead.templates.revision". That indicates that either
Kid is slow (which I think is an accepted fact) or that the template
could be restructured, e.g. at the moment the page contains both the
side-by-side and unified diffs, with one or the other being
hidden/unhidden with Javascript, for all the files in the revision.
All that must be generated before the page is returned.
It would be much quicker to render the initial page if it were
collapsed and the diffs were populated with AJAX in the background or
on demand.
Cheers,
Kent.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFGd92lgoxTFTi1P8QRAm1mAKDbC72q1Zbb2I4TfwkksWz2neb+gACfc/Jp
vAtiovTDzxOEZds+9gaTAa8=
=nWnh
-----END PGP SIGNATURE-----
More information about the bazaar
mailing list