source: trunk/third/ispell/WISHES @ 10334

Revision 10334, 4.5 KB checked in by ghudson, 27 years ago (diff)
This commit was generated by cvs2svn to compensate for changes in r10333, which included commits to RCS files with non-trunk default branches.
Line 
1Things remaining to be done to ispell:
2
3        - It might be nice to support multiple personal dictionaries.  On
4          the other hand, it's pretty easy to combine them with "cat".
5        - A small amount of string space could be saved if buildhash would
6          combine strings with common suffixes (e.g., "and" could be stored
7          as a pointer to the tail of "bland").
8        - (Pace Willisson) Pace's latest version of ispell compresses
9          common digrams to reduce the size of the hash file, and stores
10          shorter words in the dictionary entry itself.  (Since the digram
11          compression also reduces word size, this is a big win).  He also
12          improved startup time, at a slight running-time penalty, by
13          eliminating the mass conversion of string indexes to pointers
14          and just using the indexes as such whenever a string is accessed.
15        - (Peter Wan) Ispell should have a "server mode" for large sites, to
16          get rid of the time needed to read in the dictionary.  On System V,
17          this could be accomplished by having the first execution of ispell
18          read the dictionary into a shared-memory region.  Later incarnations
19          would then get the dictionary by just attaching to the region.
20          One problem would be that the dictionary gets modified during
21          the run, so you might still have to do a memory-to-memory copy
22          after the attach.  The size of having two copies of the dictionary
23          might prohibit this on many machines.  Another approach is a
24          message-based "good.c server", but this too would have to deal
25          with the possibility of modifying the dictionary.
26        - The findaffix script takes ridiculous amounts of time and disk
27          space.  It desperately needs to be rewritten in C, which would
28          also allow it to correctly support string characters and to
29          suppress reporting of choices that are already in the affix file.
30        - Some of the following ideas require more flag bits in the
31          dictionary.  Since there is only one bit remaining for most
32          cases, I plan to use that bit as some sort of an indicator
33          that more flag bits reside somewhere else.  This will be a
34          kludge, but it will save some space.  Beware!  Don't plan on
35          using that last flag bit for something else.
36        - (Ian Dall)  For some applications, it can be handy to allow
37          multiple dictionary hash files.  This shouldn't be too hard, since
38          there's already similar code to support the personal dictionary.
39        - (Not mine, but I've lost the name of the originator.)
40          Some misspellings are common, but corrections will not ever
41          be found by ispell's algorithm.  It would be nice to be able
42          to explicitly specify misspelling/correction pairs for such
43          words (e.g., "lite->light").
44        - Ispell has too much knowledge of particular text formatters
45          (i.e., nroff and TeX) wired into it.  It would be nice if
46          ispell were more flexible about this, so that it could be
47          used with some of the other file formats (e.g., Rich Text
48          Format) that are occasionally used with Unix.
49        - Several people, notably Peter Mutsaers, have asked if the
50          affix file format could be extended to allow limited
51          variables, so that you could specify things like
52          "[AEIOU][DNL] > \2ING" to handle words like "pad->padding".
53        - Ispell should be smart enough to ignore hyphenation signs,
54          such as the TeX \- hyphenation indicator.
55        - Since there can be two personal dictionaries, there should
56          be a way to specify which dictionary a new word ("I"
57          command) should be inserted into.
58        - For languages that form lots of compound words, such as
59          German, munchlist should be smart enough to split compound
60          words into their components when appropriate.
61        - (Jeff Edmonds) The personal dictionary should be able to
62          remove certain words from the master dictionary, so that
63          obscure words like "wether" wouldn't mask favorite typos.
64        - (Jeff Edmonds) It would be wonderful if ispell could correct
65          inserted spaces such as "th e" for "the" or even "can not"
66          for "cannot".
67        - Since ispell has dictionaries available to it, it is
68          conceivable that it could automatically determine the
69          language of a particular file by choosing the dictionary
70          that produced the fewest spelling errors on the first few
71          lines.
72        - It is long past the time when the ispell.1 manual page
73          should have been broken up into components describing the
74          various programs in the suite.
75        - If the -C flag is disabled, ispell should (at least
76          optionally) use the "??" form to suggestion possible
77          compound formations.
78        - The elisp interface should provide a way for ispell to
79          return error messages to emacs, so that users don't get
80          inexplicable failures when things like dictionary open
81          failures happen.
Note: See TracBrowser for help on using the repository browser.