Re: LYNX-DEV -color/-nocolor development confusion (was: fotemods.zip update)

[Foteos Macrides]

Klaus Weide <address@hidden> wrote:
>On Mon, 21 Jul 1997, Foteos Macrides wrote:
>>      An update of fotemods.zip is available in:
>> 
>>      http://www.slcc.edu/lynx/fote/patches/
>> 
>> 1997-07-21
>> * Added the "show color (&)" binary (ON/OFF) choice in the 'o'ption menu.
>>   It will be present if color is supported, and the setting can be saved
>                                                              ^^^^^^^^^^^^
>>   to the .lynxrc file.  Once saved, it will be used as the startup default,
>    ^^^^^^^^^^^^^^                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>   but if OFF it can be overridden via the -color command line switch or
>>   the COLORTERM environment variable.  Also, updated the online 'h'elp
>>   accordingly. - FM
>
>Very bad idea.
>
>The setting "can be saved" == "it *will* be saved" when a user with this code
>presses '>' on the Options Screen - whether the user has changed it or not,
>whether the user is aware of it or not, whether it is appropriate or not.
>
>This may help some who otherwise have no way of setting a terminal
>type or command line flag to control use of color, *and* who always
>uses the same kind of terminal.  It messes things up for all those who
>are using terminal types as they should be used, and have the curses
>(or slang) system automatically decide on startup whether to use
>color.
>
>Apparently you have taken part of Tom's code and/or user interface
>(which he wrote mostly for color-curses in mind) and "adapted" it to
>work "better" with slang in the fotemods code.  Well in some respects
>it works better, in others it is worse.
>
>Neither your code nor devel ac-0.38 always show the current status
>(whether color or not) correctly on the Options Screen.
>
>There is some flaw in the logic of the FM code when determining the
>current state of the variable for the Options Screen, and/or in the
>code which determines whether the screen needs to be refreshed
>after a (presumed) change:
>   #                    No saved color setting in .lynxrc
>   unset COLORTERM
>   TERM=linux lynx .  # a TERM which supports color
>   'O'ptions          # show color falsely shown as OFF (!)
>   '&'
>   RETURN             # immediately, without change
>                      # show color still falsely shown as OFF
>   '&'                # again
>   RETURN             # again immediately, without change
>After this, color use is turned off but without a screen refresh.
>It makes a mess of the screen if colors with non-default bachgrounds
>(the default colors) are used.
>
>You have changed the function of that new Options Screen setting
>(and of the LYShowColor variable) significantly: in Tom's code it only
>applies for his color-curses, not for slang color.  In Tom's code it
>can only be used to force color OFF (against what terminfo says).  In
>your implementation it can force color ON or OFF, against what
>terminfo says.  I think that is a resonable extension, given that
>slang-color can do that while the curses-color implementation cannot.
>But Tom and you should be aware of the different semantics when
>comparing the code.
>
>What I would like to see for the devel code:
>
>- NO saving of LYShowColor in .lynxrc.
>  [currently: FM-no, TD-ok]
>
>- A -nocolor flag that works for both slang and curses.
>  [currently: FM-no, TD-for curses only]
>
>- Continued support for -color and COLORTERM for slang (as it is now).
>  It doesn't make sense for curses color since there are no default
>  ANSI-ish color escapes built in, so one cannot force color on if the
>  terminal type says "no way".
>  [currently: FM-ok, TD-ok]
>
>- Possibility to toggle color on and off on the Options Screen, for
>  the current session only (like the User-Agent string), if color
>  is compiled in.  
>  [currently: FM-slang only and will be saved, TD-for curses only]
>
>- The setting shown on Options Screen should always show the state
>  currently in effect (except while it's being changed, of course).
>  I.e. it is not just an additional requirement for <Is-color-being-used>
>  (as in Tom's code), it is <Is-color-being-used> itself (as apparently 
>  intended by fotemods).
>  [currently: FM-no, TD-no]
>
>- Options screen toggle should work the same for slang and curses color
>  support, as far as possible.  The reservation means that for curses
>  color, one cannot turn color on if the terminal description says that
>  the terminal doesn't support color.
>  [currently: FM-no, TD-no]
>
>- Trying to change the color setting on the Options Screen when it is
>  not possible (will have no effect) should result in an appropriate
>  statusline message.
>  [currently: FM-no, TD-no]
>
>"Currently" refers to fotemods of 1997-07-21 for FM, to code 2.7.1ac-0.38
>for TD.
>
>Tom, please don't just copy Fote's for-slang code without checking.
>We can create our own bugs just fine :) :).

        The fotemods code presently offers internal color support only
via the slang library.  The "show color (&)" option appears in the
'o'ptions menu only if the image was built with slang.  It can be
saved to a .lynxrc file only if the image was built with slang, and
the account is not anonymous.  It no show_color:on is in the .lynxrc
file, and the image was built with slang, it comes up off unless
the -color switch was used or the COLORTERM enviroment variable is
set.  If show_color:off is in the .lynxrc file, that's the default
anyway, and will be overridden if the -color switch is used or the
COLORTERM enviroment variable is set.  To the extent I've exercized
yesterday's fotemods, when I use an image built with slang and invoke
the 'o'ption menu the "show color (&)" value (ON or OFF) accurately
reflects whether it's on or off, and does toggle that successfully.
Most of your claims above are false.  For example, when invoked in
an anonymous account, '>' is a dead key, and information on how to
save options is omitted from the instruction line.  We are all
capable of creating our own bugs just fine, and if there really are
any bugs in yesterday's fotemods, hopefully I'll detect them or someone
will report them, so that they can be corrected.

        I'd been checking all morning and afternoon for -38 to show
up at the update page, and finally it has.  I built it on Scott's
solaris, and the statusline glitch when using a non-color terminal
appears to be corrected.  However, when connected via a v220 emulator,
if I invoke the 'o'ptions menu the "show color (&)" claims ON, though
it should be OFF, and when I toggle it, nothing seems to happen. (I
hope this bug report is not lacking. :).

                                Fote

=========================================================================
 Foteos Macrides            Worcester Foundation for Biomedical Research
 address@hidden         222 Maple Avenue, Shrewsbury, MA 01545
=========================================================================
;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;