lynx-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: LYNX-DEV multiple bookmark support and popup/radion_button in 'o'pti


From: Foteos Macrides
Subject: Re: LYNX-DEV multiple bookmark support and popup/radion_button in 'o'ptions. 0.09
Date: Wed, 27 Nov 1996 17:07:27 -0500 (EST)

"Hiram Lester, Jr." <address@hidden> wrote:
>On Wed, 27 Nov 1996, Filip M Gieszczykiewicz wrote:
>
>> Greetings. Eeek.. I forgot to upload the patch for that. So sorry.
>> It's up to 0.09 and the key fix is line #521 or so, fix up the test to
>> be:
>
>If Fote doesn't get around to it, I'll try to get it in over the holidays.
>
>> Problem is array range.. c gets 'A' subtracted from it by c = TOUPPER(c) - 
>> 'A'
>> and then becomes the index for an array... (bonehead bug comes in) 13-'A'
>> is a negative number... and WHAMO!
>
>That could be a very bad thing. :)  Again, I see HP-UX is more tolerant
>and doesn't core dump.  Even Linux wasn't core dumping only crashing.
>
>> > Also, one other thing that should probably be cleaned up is that the
>> > second screen of bookmark files when editing the files is shifted one line
>> > down from the first screen, and it looks slightly awkward when flipping
>> > back and forth between screens.
>> 
>> Geez :-) I thought I'd get away with that. You know, I wrote it and I
>> can't track it down (why it happens). I think the part that decides
>> how many lines to show per screen has some int-division problems.
>> 
>> It's really noticable when you have a 9 row display (I tried), first screen
>> has 3 entries, second and on ONLY 2...  Shoot me! :-)
>
>I'll try to take a look at it this weekend if Fote doesn't get it before I
>do. :)
>
>I also discovered another oddity probably left over from the single
>bookmark file code.  When you are viewing bookmark file A and add a link
>from it to another bookmark file, you cannot 'r'emove the bookmark from
>file A without 'v'iewing file A again.  This was probably in the original
>code to prevent you from adding a bookmark and trying to remove one
>without reloading the page.  Reloading won't help this one either.  This
>behavior might seem normal if you were adding a bookmark to the current
>bookmark file, but not to a different file...

        I started out with Fil's patch ;( which unfortunately was 07,
because Fil didn't replace it with 09 ); so it would be like what
people who were using that had become accustomed to, but it was
crashing left and right when I tried it, wasn't taking all kinds of
things about security, and factors associated with 'a'dding or 'r'emoving
links across bookkmarks, into account, etc.  So I was modifying it
literally for several days before putting it out for further testing
by others, and don't think it should be in the Composite yet (didn't
think you'd be updating that this quickly 8-).  I've been doing more
fixes and mods of it today, which I'll put out tonight.

        I encountered that "line #521" bug right off the bat (i.e.,
it crashed as soon as I tried it with the original patch 8-), and what
I put out yesterday doesn't crash with any of the exercising I've done
so far, so try to figure out where it's happening for you, Lester, in
case the problem is still there in the code I put out tonight, but
doesn't show up here.

        In any case, though the look and feel of it is the same as with
Fil's patch, the underlying code at this point is considerably different
in my mods set (and still changing 8-).

        I'm going to get rid of that prompting instead of showing the
multiple bookmark menu in ADVANCED mode.  If you don't have multiple
bookmarks, just turn the feature off via the 'o'ptions menu, and if
you do have them, the menu scheme is now efficient enough that I'd
just as soon see it each time, as with NOVICE or INTERMEDIATE mode.
I didn't change the sizing logic, though, so that round-off variation
is still there. 

        In any case, use your own judgement, Hiram, in what you put
into the Composite, and I'll let you know when the version here is
stable enough to include, if you want that.  Perhaps Fil might want
to look at it too (tomorrow, so it has the new stuff I'm doing today),
for comparison.

        I think it was a good start on Fil's part, but even being
intimately familiar with all of the Lynx code, I'm having a hard time
making everything that needs to be taken into account actually taken
into account.  Once I've done that, I'll then work on making it yet
more efficient.

        I guess 'cuz Rob hasn't updated LynxRP since October, it's
natural for the Composite to start being treated as the "development"
code set, and I certainly don't want to discourage Hiram from actually
doing code development instead of just consolidating the stable stuff,
but could the Composite stay the "clear bug fixes and really stable
enhancements for both Unix and VMS" (nothing is ever "perfect" of
course 8-), and perhaps Hiram could figure out a way to overlay "needs
more testing but builds on both Unix and VMS" stuff on top of that
(e.g, start up a Hiram's mods patch or replacement set, which similarly
gets added to the Composite when it's been field tested enough to appear
stable)?  That's just a suggestion, and if it's too complicated, forget
it. :) :) :)

        I think it's really important, though, in the long run, for
people to do some "back and forthing" themselves about their mods,
or we'll never again have a reasonably large group of people who
are reasonably knowedgeabe about the actual Lynx API and code, and
not just proposing more "angels" that "somebody" should fit onto the
head of that "pin". :) :) :)
                                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.
;

reply via email to

[Prev in Thread] Current Thread [Next in Thread]