Tech Arsenal 1/Tech Arsenal (Arsenal Computer).ISO/tek-13/pcvile.zip/BUGLIST

Text File | 1991-06-20 | 6KB | 148 lines

    (E means enhancement, L,M,H are low, medium, high priority)
----------------------

H    the gotoeop motion and regions interract wrongly.  For instance,
    using ^W} to write the rest of the paragraph to a file writes
    the wrong number of characters.

H    when reading the output of a shell command, it's far too slow -- should
    read as much as possible, put it into the buffer, and _then_ update
    [ This has been fixed for BSD -- other implementations not yet ]

M    output doesn't flush properly after a shell escape

M    longnames cause severe portability constraint -- does the shortnames
    stuff take care of this?  I can't test it...

M    :k, to set a mark, won't work as ":ka" or ":kb".  Must use ":k a"

M    :set all should be made to work (fix help)

L    formatregion sometimes appends a space to the last line, for instance
    when a paragraph is reformatted

L    dos-style is set even when file is majority unix-lines?

L    why does file reading seem so slow?

L    why is searching so slow?

L    with two window displayed, hitting * twice makes one go away.
    perhaps popup should _always_ split a window, rather than sometimes
    taking over an existing one.  but then we'd have to remember who we
    split, so we could give the space back to the donor.

E    patterns as addresses do not work, e.g. ":/str1/,/str2/d".  They're
    hard to parse the way things are set up right now.  We could accumulate
    the whole commandline, and then parse it, the way real vi does, but we'd
    lose the "prompt and display last response" behavior.

E    In vi, the join command is supposed to act either on a region (from
    the command line, as in ":13,15j"), or it should take a simple
    count, from vi mode, as in "3j".  Right now vile _only_ does the
    simple count form of the command.

E    should add an option to support file locking, rather than the
    current ifdef stuff -- but this is only useful if we match the
    GNU locking protocol, which I'm not inclined to do

E    Wow. the scrsearch functions could become region based -- as in "search for
    the next occrence of the region", which would usually be a word.  And
    the ^A/ version could become "a/ (search for teh contents of buffer a),
    if you know what I mean.

E    need a "file newer than buffer" warning.  Should stat the file, and
    store it's mod time, compare to current when going back to that window,
    after performing some shell command.  e.g., if you're editing a file
    that is created by a script you're testing, then you want to be warned
    that the file is out of date after doing a test run of the script.

E    should be able to refer to remote tags file, and have paths
    contained there be relative to _its_ location, rather than the current
    directory.

E    add a command-line expansion character to be the list of files
    mentioned in "tags"

E    should there be a "significant-length" for tags?

E    another mode, which will "!get -p" an SCCS or RCS file if
    the clear-text file isn't found

E    add _another_ mode, similar to the above, which will do a caseless
    match against filenames mentioned in tags

E    it would be nice if ^X!! reran the last command into [Output]

E    add support for sh or C comments to formatregion() code

E    why can't I kill a displayed buffer?  simply bring up the previous.

E    g should become a region command.  Then it could take ranges, as
    it should, and could also become an operator command.

E    add C comments to CFENCE matching code

E    add C ifdefs to CFENCE matching code

E    adjust window size of popups based on length of buffer.  currently
    popups get half the window they're splitting, no matter what

E    collapse command execution code to as few places as possible.
    Its currently spread through main(), execute(), operator(),
    docmd(), and usekreg().

E    mlreply line should ideally be a one line buffer, so editing
    and history can be done on it.

E    BSD interrupt processing is botched during a read() of the keyboard.
    The read doesn't return -1 as it does under sysV (USG).
    So you can bang on ^C all day and nothing will happen.
    [I'm not sure this is true anymore  -pgf]

E    This editor does not virtualize buffers out to disk.  It is quite
    a memory hog.  But isn't that what /dev/swap is for?  But
    seriously, I haven't even come close to testing it for
    memory-full conditions.  Some malloc() packages give 95%
    warnings -- perhaps something like that should be done for
    safety.

E    to make the code shrink, could probably try to eliminate
    sprintf and its brethren.  mlwrite already does format
    expansion, it could probably be turned into a doprnt-like
    beast, and that used as a basis for our own sprintf,fprintf
    How does one write [sfp]printf and doprnt using varargs?

E    likewise for scanf

E    marks should perhaps be linked onto lines.  this would make a lot
        of things a lot easier, since a mark would travel with the
        line, instead of having to be moved when the line is
        reallocated etc.  the U line could be treated as a special
        mark.  The "copied" flag needed by undo could be a special
        sort of mark as well.  Implementation of the "tag stack"
    would be aided by this as well.

E    there is currently no way to change working directories -- it's
    not clear whether that should be a global thing, or perhaps
    a per-buffer notion.

E    there is code (in imdying()) that attempts to save unwritten buffers
    on hangups.  It is fairly crude, having been written quickly
    in self-defense during debugging.  It should be examined and
    fixed.

E    :e and :n should be able to deal with multiple filenames resulting
    from filename globbing.  vi has this problem too.  At least
    vile lets you choose to choose the first such name.  if should show
    you the first name, so you know whether to accept it or not.

E    ^W should erase word in input mode, and killc should erase line
    in input mode.  (but only back to where the input started, if it
    started on this line.  right now we have no record of that, hence
    you can backspace past the insert point)

E    All code should pass lint.  This won't be easy... :-)