id-utils TODO - Future directions

* General
  - Add comments to all substantive functions

* Portability
  - Fix file name handling for non-UNIX file systems
    (notably DOS, Windoze & OS/2).
  - Fix assumption that (sizeof (int) == sizeof (long))
  - Integrate changes for MSDOS via djgpp
  - Port to cygwin
  - configure gcc warning flags
  - remove arbitrary buffer-size limits and unsafe libc functions (e.g., gets)

* lid
  - use "long long" (if available) integer comparisons
    in query_number ().

* mkid & lid
  - store & retrieve floating point literals
  - automatically crack (optionally gzipped or compressed) tar files, so
    you can index and search archived source without having to unpack it.
  - reset access times

* mkid
  - (Optinally?) store tokenized contents of comments.
  - Store attributes with symbols: type, func, variable struct-tag
    enum-tag, keyword, etc.  Store caller/callee relationships between
    identifiers.  Indicate which files have defns vs. uses.  This info
    is needed to support a cscope interface.
  - Add times(2) results to statistics output.
  - Scale statistics base unit (KB, MB, GB)
  - treat \ specially for TeX files and do likewise with @ for Texinfo.
  - report dangling symlinks as such, rather than just saying "can't stat"
  - don't gripe about dangling symlinks if file won't be scanned anyway
  - Reinstate "-" on command-line, meaning read stdin for newline-separated args.
  - detect and avoid cycles in the tree induced by symlinks to /.

* gscope
  - add a cscope work-alike query interface
  - rewrite scanners to be scanner/parsers that can distinguish defn
    from use, lvalues from rvalues, types from vars from funcs from macros.

* grep
  - Hack GNU grep to use a mkid database for hints about which files to
    search.


Local Variables:
mode: text
fill-column: 75
End: