summaryrefslogtreecommitdiff
path: root/TODO
diff options
context:
space:
mode:
authorTor Andersson <tor@ghostscript.com>2004-11-17 08:41:37 +0100
committerTor Andersson <tor@ghostscript.com>2004-11-17 08:41:37 +0100
commitfc5d174cd475c8f2236ea644d63e6b309e41b928 (patch)
tree72dec2aa55a40b448547a2db77055bb02a172bdd /TODO
parented0f143ca6508817001319c6668602a07fe421dd (diff)
downloadmupdf-fc5d174cd475c8f2236ea644d63e6b309e41b928.tar.xz
text clip mode. split text content streams. text extraction api.
Diffstat (limited to 'TODO')
-rw-r--r--TODO20
1 files changed, 9 insertions, 11 deletions
diff --git a/TODO b/TODO
index 5be19c71..bfeaee87 100644
--- a/TODO
+++ b/TODO
@@ -1,37 +1,36 @@
-colorspace conversions
+colorspace conversions (v2)
- cal*
- iccbased
- how to normalize Lab components to 0..1
- fast color cubes
- how to cache colorspace cubes (what key?)
-image rendering
+image rendering (v2)
- tiles
- dct case
- better filter than box
+ - lazy decoding
shadings
- ... jeong ...
rendering
- - save non-transformed bbox in nodes
- - explicit mask field in fz_renderer .. general cleanup
+ - bbox culling (cache bbox in over node?)
+ - image mask + color case
- merge gka optims
- - optimize! optimize! optimize! (special case 1 and 4 channel cases)
+ - optimize inner rendering loops
+ - optimize image load/decode/scale
+ - special-case optims (1, 2 and 4)
+ - cpu-specific optims
parser
- - text clip mode
- - split content streams (TJ objects)
- resource dict generate fake ids
- try to clean up colorspace/material handling in interpreter
- - tounicode
clean up
- make source ansi c89 / pedantic
- reference count everything
- standard cleanup mechanism
- - naming conventions (fz_new/renew)
-
- design by contract
- split into private and public
- comments and documentation
@@ -39,5 +38,4 @@ clean up
cache
global cache for cmaps and fontfiles (emb+sys)
render cache (link-nodes and scaled images)
- profile font cache (esp with t3 fonts)