summaryrefslogtreecommitdiff
path: root/dviware/catdvi/TODO
blob: f753663ff51ac8b9434d2b22e0963fa08cc5a38c (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
IMPORTANT                            -*- Text -*- 

  * Figure out how font's space params relate to the whitespace between
    words in DVI.
    [the current heuristics seem to work quite well however]

  * Go through the current font encoding translation tables and fix the
    unknown glyphs (do a make enc/*.h after ./configure
    --enable-developer-warnings to see the problems).
    [nearly done]

BUGS:

    [None except the known limitations]

SMALL IMPROVEMENTS:

  * Add encodings: TEX extended ASCII.
  * should ./configure --enable-developer-warnings add
    -Wall etc. to CFLAGS if we're using gcc?

CLEANUPS

  * page.c :
     - box_t has grown large. Do we need a constructor?
     - Put diacritics, texmext, radicals in a list and loop over the list
       to adjust them, not over the whole page.
  * fntenc.c: add additional per-encoding info, like "this encoding has the
    same params array as OMS".
  * Rethink the *clean targets in the makefile.
  * Reorganize the glyph tweak tables in outenc.c
  * Referencing fonts by their number begins to get ugly. Switch to pointers.

PROJECTS

  * Faciliate recovery of lost LaTeX files: introduce a "LATEX" output
    encoding which spits out LaTeX commands for at least the most common
    symbols. (Thanks to Giovanni Salme for suggesting this).
 
  * Supplements to outenc.c :
      - Add more visual substitutions for ascii and latin1 output encodings:
        Arrows, angle brackets, lessmuch, greatermuch, ...

  * Extend or mimick the diacritics mechanism to glue "multipart" glyphs
    (long arrows, mapsto, hookarrow, etc.) together. The same could also
    work for extended recipies.

  * Math layout improvements -- the word break heuristics we use work well
    with normal text, but fail miserably with math (no wonder since
    TeX uses different spacing methods for math and the document author
    usually adds another, impredictable, bunch of spacing instructions).
    This sometimes leads to funny spacing in catdvis output.

    One possible approach would be: there are no word breaks in formulae.
    Position every glyph in a formula based on its DVI coordinates (as we do
    with word beginnings). Think about how to pick the glyphs that we have to
    position directly. Possibilities:
     - based on glyph hints (i.e. per-unicode-char information)
     - based on per-font information (math chars form extra character sets,
       at least with TeX). Not really nice.

  * Add rule support. Most important are line-like rules for fractions,
    tables, etc. A lot of the required infrastructure is now implemented
    (canvas.c and layout.c). Still to be done:
      - Reserve sufficient space to print the rules, but allow rules to
	intersect (for table borders etc.).
    Most likely, special private space glyphs will be inserted into the page
    to represent the rules. If so, then
      - Try to detect fraction slashes and treat them like "on axis" texmext
	glyphs.
      - Define some kind of bogus font metrics for these glyphs.
      - We will need "subclasses" of page_word_t to make this work.
      - And some special case "break here" test in the word breaking code.
	Maybe make this generic and combine with better math spacing attempts.

  * What about a Win32/MikTeX port? Current MikTeX seems to include libkpathsea.
  * Think about the pros and cons of using GNU automake.

WISHLIST

  * Implement TFM file location routine also without kpathsea
  * Add support for non-latin languages. Cyrillic should be easiest (but
    will need native speakers to verify the implementation), right-to-left
    languages a lot harder.
  * Looking at xdvik and kdvi, it may be time for a catdvi library.