Thorsten Haude wrote:

>  1. Determine release-critical bugs 
>     Meaning we should all huddle and create a list of things that have to
>     be fixed before RC1. This would include stuff from the SF tracker but
>     also anything else we could come up with.

Well, although the quote refers to bugs and also contradiciting my
previous post, there are two issues that come to my mind:

 1) In the new release the command shell logic will change: Instead of
    having a hardcoded default value (ie, /bin/csh), the user's login
    shell is chosen.  Moreover, the shell setting was moved to the GUI
    (preference settings), so it's easier to change.
    
    Now, if you use or share macros that make use of shell_command(),
    you can't rely on a fixed default value for the shell. (Although
    the shell could be changed before via an X resource setting, many
    users were not aware that nedit was using csh, unless they found
    out about how to set the shell.)  Since the command syntax differs
    from shell to shell, macros using the shell need *at least* to
    know which shell is actually in use, so this calls for a
    $command_shell variable.  All the other preference settings are
    also available in macros having a variable and even an action
    to set them, see Window Preferences Actions (under Action Routines).
    Would be even better to have both these things for the shell, too, 
    so one could write
        
        old_shell = $command_shell
        set_command_shell("/bin/csh")
        shell_command(...)  # with C shell syntax in this case
        set_command_shell(old_shell)
        
 2) For 5.4 background coloring was made available for syntax
    highlighting.  This coloring stops at \n. Then Tony found out how
    to make the coloring go beyond \n and asked for comments (ie,
    include into 5.4 or not).  Unfortunately, back then the change was
    misunderstood as an either-or decision (either stop at \n or go
    beyond \n always), which is wrong, because now we have "always
    stop at \n", and with the change we would have both possibilities
    "go beyond \n" and "stop at \n" (achieved with a subpattern
    catching the \ns).  
    If anyone had been aware of this back then, this change had surely
    gone into 5.4. Therefore, I wonder if the unfortunate omission
    shouldn't be corrected eventually:
    
    [four lines to be changed in textDisp.c:2184]
    
    if (lineIndex >= lineLen)
        style = FILL_MASK;
    if (styleBuf != NULL && textD->highlightCBArg != NULL) {
        /* pick up styleBuf's style value separately */
        int styleChar = (unsigned char)BufGetCharacter(styleBuf, pos);
        if (styleChar == textD->unfinishedStyle) {
            /* encountered "unfinished" style, trigger parsing */
            (textD->unfinishedHighlightCB)(textD, pos, textD->highlightCBArg);
            styleChar = (unsigned char)BufGetCharacter(styleBuf, pos);
        }
        style |= styleChar; /* add styleBuf value found to style
        bitmap */
    }
    

Jörg
--
NEdit Develop mailing list - [email protected]
http://www.nedit.org/mailman/listinfo/develop

Reply via email to