1
- Go through things marked FIXME
3
- Add calls to prepare and finish access where necessary. grep for
4
ACCESS_MEM, and make sure they are correctly wrapped in prepare
7
- restore READ/WRITE in the fbcompose combiners since they sometimes
8
store directly to destination drawables.
10
- It probably makes sense to move the more strange X region API
11
into pixman as well, but guarded with PIXMAN_XORG_COMPATIBILITY
13
- Reinstate the FbBits typedef? At the moment we don't
14
even have the FbBits type; we just use uint32_t everywhere.
16
Keith says in bug 2335:
18
The 64-bit code in fb (pixman) is probably broken; it hasn't been
19
used in quite some time as PCI (and AGP) is 32-bits wide, so
20
doing things 64-bits at a time is a net loss. To quickly fix
21
this, I suggest just using 32-bit datatypes by setting
22
IC_SHIFT to 5 for all machines.
24
- Consider whether calling regions region16 is really such a great
25
idea Vlad wants 32 bit regions for Cairo. This will break X server
26
ABI, but should otherwise be mostly harmless, though a
27
pixman_region_get_boxes16() may be useful.
29
- Make source clipping optional.
30
- done: source clipping happens through an indirection.
31
still needs to make the indirection settable. (And call it
34
- Consider optimizing the 8/16 bit solid fills in pixman-util.c by
35
storing more than one value at a time.
37
- Add an image cache to prevent excessive malloc/free. Note that pixman
38
needs to be thread safe when used from cairo.
40
- Review the pixman_format_code_t enum to make sure it will support
41
future formats. Some formats we will probably need:
43
ARGB/ABGR with 16/32/64 bit integer/floating channels
47
Also we may need the ability to distinguish between PICT_c8 and
48
PICT_x4c4. (This could be done by interpreting the A channel as
49
the depth for TYPE_COLOR and TYPE_GRAY formats).
51
A possibility may be to reserve the two top bits and make them
52
encode "number of places to shift the channel widths given" Since
53
these bits are 00 at the moment everything will continue to work,
54
but these additional widths will be allowed:
56
All even widths between 18-32
57
All multiples of four widths between 33 and 64
58
All multiples of eight between 64 and 128
60
This means things like r21g22b21 won't work - is that worth
61
worrying about? I don't think so. And of course the bpp field
62
can't handle a depth of over 256, so > 64 bit channels arent'
63
really all that useful.
65
We could reserve one extra bit to indicate floating point, but
68
PIXMAN_TYPE_ARGB_FLOAT
69
PIXMAN_TYPE_BGRA_FLOAT
72
image types. With five bits we can support up to 32 different
73
format types, which should be enough for everybody, even if we
74
decide to support all the various video formats here:
76
http://www.fourcc.org/yuv.php
78
It may make sense to have a PIXMAN_TYPE_YUV, and then use the
79
channel bits to specify the exact subtype.
81
What about color spaces such a linear vs. srGB etc.?
86
- Run cairo test suite; fix bugs
87
- one bug in source-scale-clip
89
- Remove the warning suppression in the ACCESS_MEM macro and fix the
90
warnings that are real
93
- make the wrapper functions global instead of image specific
94
- this won't work since pixman is linked to both fb and wfb
96
- Add non-mmx solid fill
98
- Make sure the endian-ness macros are defined correctly.
100
- The rectangles in a region probably shouldn't be returned const as
101
the X server will be changing them.
103
- Right now we _always_ have a clip region, which is empty by default.
104
Why does this work at all? It probably doesn't. The server
105
distinguishes two cases, one where nothing is clipped (CT_NONE), and
106
one where there is a clip region (CT_REGION).
108
- Default clip region should be the full image
110
- Test if pseudo color still works. It does, but it also shows that
111
copying a pixman_indexed_t on every composite operation is not
112
going to fly. So, for now set_indexed() does not copy the
115
Also just the malloc() to allocate a pixman image shows up pretty
120
- Make all the setters not copy their arguments
122
- Possibly combined with going back to the stack allocated
123
approach that we already use for regions.
125
- Keep a cached pixman_image_t around for every picture. It would
126
have to be kept uptodate every time something changes about the
129
- Break the X server ABI and simply have the relevant parameter
130
stored in the pixman image. This would have the additional benefits
133
- We can get rid of the annoying repeat field which is duplicated
136
- We can use pixman_color_t and pixman_gradient_stop_t
137
etc. instead of the types that are defined in
added
removed
TODO
