[cairo] Call for unresolved issues (and planning for 1.4)

Carl Worth cworth at cworth.org
Sat Jan 20 01:42:53 PST 2007

As some of you may have noticed, I've been rummaging through the cairo
mailing list from the past few months looking for loose ends that
could be tied off (things like dropped patches and particularly nasty
bug reports).

At this point, I've tried to address everything that looked obvious to
me, and either applied the dropped patch, fixed the bug, or apologized
and asked for more help.

But I'm sure I've missed some things, so if you've sent a patch that
seems to have been forgotten, or you've reported a particularly nasty
bug that seems to have been neglected, please bring that to our
attention, (often replying to the original message will work better
than starting a new thread).

So I'm going to go ahead and push out a 1.3.12 snapshot with what's in
place now. There are 3 notable bugs that I wish we would have fixed,
but I've added them to the ROADMAP file for the next snapshot as I
plan to give these attention in the next week. These are:

   "issue with rotated image sources" from Benjamin Otte

   "Problem with linear gradients and reflect mode post-rewrite" from
   T. Rowley

   #8801: text rendering lacking locking in multithreaded apps

And for each of these, I've just posted something to the mailing list.

Meanwhile, it's time we start working on finishing up the 1.4.0
release. We've made a lot of great performance improvements during the
1.3.x series, and I'd really like to get them out in a major release
so people can stop using that old, slow 1.2 series.

I'd be interested in hearing what people would definitely like to see
still happen before 1.4.0. Obviously, we're not done with performance
work, and that will continue after 1.4, (for example, I'd love to see
a re-architecting of pixman with more automated compilation of
code---and I think we've now chased a lot of performance problems into
the X server where they will need some further attention). So, what
I'm looking for is smaller performance fixes, or non-performance-
related items that should be fixed before we call 1.4 complete. Here
are some ideas:

    * Fix selection of text in PDF output (Alp has been threatening us
      with a patch for this)

    * Fix PDF to not do image fallbacks due to gradients, (Miklós made
      a patch, and I've made some comments more recently)

    * Implement a faster trapezoid rasterizer (I've got a plan for
      this that should be easy---I've just been waiting for a test
      case that really shows rasterization as a bottleneck to justify

    * Resolve the various new "list getters" API (see current thread
      debating the multiple approaches embodied in the new API calls)

    * We should probably workaround this nasty X server bug:

      backend-content                 test-size min(ticks) min(ms) median(ms) stddev. iterations
        image-rgba    long-lines-uncropped-100    8522618    5.529      5.541  0.70%  20
         xlib-rgba    long-lines-uncropped-100  906578296  588.125    589.172  0.15%  5

      Yes, you're reading correctly that the X server is drawing
      something 500 times slower than cairo's image backend. The X
      server fix for this is pretty simple, (just need to not allocate
      temporary trapezoid mask surface any larger than destination
      surface). But we should fix cairo to make this fast even for
      unpatched servers, (easiest will be to just allocate the mask

    * Fix some long-standing bugs, (like EXTEND_PAD for surface
      patterns), perhaps other tests showing expected failures as

    * Maybe throw in some blurring support through
      CAIRO_FILTER_GAUSSIAN? (even before we establish more
      sophisticated means of controlling Gaussian blurring or other
      filter types?)

Anyone have anything else they don't want to see get left out of 1.4?

I'd like to see 1.4 happen very soon, (we've been saying January 2007
in ROADMAP, and suddenly that's almost as close as the next weekly
snapshot (yikes!)).

But the list above is still pretty long---and definitely more than one
person can do in a week. So we're going to need a lot of us working
together to make this happen, (or let some things slip). Let's
everybody that can put some great work into cairo over the next week!

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.freedesktop.org/archives/cairo/attachments/20070120/224af83a/attachment.pgp

More information about the cairo mailing list