Re: LYNX-DEV Re: Lynx updates

[Klaus Weide]

[For the impatient who just want to know "does that affect me", see
close to the end."]

On Sat, 9 Nov 1996, Foteos Macrides wrote:

> "Hiram Lester, Jr." <address@hidden> wrote to me an Rob instead of
>  lynx_dev:
> >On Wed, 6 Nov 1996, Foteos Macrides wrote:
> >
> >>    When you get to it this weekend, just discard the stuff I had
> >> emailed to you and get the mods zip in http://www.wfbr.edu/dir/lynx
> >> with my suggestions for the Composite as of whenever you get it.  Note
> >> at this point it includes everthing that's been posted or set up as
> >> links at NYU, except the old patches for lynx2-5[FM], Klaus' substantive
> >> DIRED_SUPPORT mods, the Multibookmark stuff, and the actual SSL hooks
> >> (nor patch 11, mouse stuff, and color/style "new development" stuff).
> >
[ SNIP ]
> >
> >>    Klaus' mods look good, except that when the list is created,
> >> a function for freeing everything is needed, and should be passed
> >> to an atexit().  Look at any of the other list creations as models.
> >
> >Ok, after yours and Klaus' discussion on the list about him doing
> >something equivalent, should this go in? 

I believe the atexit() question has been settled.  

> > (I saw you included one thing
> >about nlinks in your mods.)  I never use DIRED and know next to nothing
> >about it... :(
> 
>       The nlinks fix is an old, obvious patch.

[ and doesn't really have much to do with the DIRED fixes.  I just
included it in that packet, because it had at that time not been
incorporated anywhere on the fixes & patches Web pages. ]

> I'm not in a position
> to excercise the more extensive DIRED_SUPPORT mods 

I believe that if you had exercised the DIRED stuff, the code would
not have been in the broken state I found it in ;)  [quite a difference
from the bulk of Lynx code which in general is well tested, an works.]

> on and adequate range
> of flavors, so it should be kept up by Klaus 

ok I can do that, but...

> until more people have tried
> it and it appears OK for the bug-fix and obviously OK replacement file
> sets.  Has anybody tried it?

Yes, I have :)

I am afraid that, if we are gonna wait for a quorum of people to confirm
that these patches are harmless, it's gonna be a long wait.  These days
it appears hardly anybody on this list wants to admit to knowing what
DIRED is, let alone _using_ it.

But there were bugs.  The worst one IMO was in the screenful or two
of lines in LYMainLoop.c that (try to) manage the list of tagged 
items. (starting around L2370 in the "official" Lynx version.) 
I think anybody with understanding of some C, malloc(), and Lynx's FREE()
macro should be able to see that it couldn't work the way it was coded
(without knowing anything about DIRED mode, except that that piece
of code was attempting to add items to and remove items from a linked
list).

I seem to be blessed with an environment (version of libc, or whatever)
or that notices easily when the code tries to access invalid memory
(that has already been freed) - a few hits of the 't' key in DIRED mode,
and oops there's that "A Fatal error has occurred" again.  [That was the
previous state - not any more with the fixes.] 

People with more forgiving environments wouldn't notice immediately,
or would just notice that the 't'ag key doesn't work as they think it
should and move on. (And then maybe get a fatal error later in the
same session, without relating it in any way to DIRED mode).  Well I
call that a Bad Bug, and think that such should not be left in the
code.  Changing the list handling to use the HTList_* functions
(proven to work, since they are used everywhere) necessitated many of
the extensive changes in LYLocal.c.  Other changes were rather
cosmetic in comparison.  [Making things work as they were supposed to,
when they don't work or don't work properly, _is_ minor in comparison
to a Bad Bug...]  Note that nearly everything of the extensive changes
is within `#ifdef DIRED_SUPPORT' sections.  

In summary, bugs were reported, fixes were provided, they are largely
restricted to features of Lynx that (1) were broken, (2) can be easily
disabled, (3) are, apparently, only used by few.  Either the fixes
should be added soon to "the bug-fix and obviously OK" replacement
set, or maybe it's time to drop that whole sack of problems from the
Lynx code (together with WAIS and other nice-but-nobody-really-cares
features).  <RANT> Then, <STRONG>and with the Lynx-dev archives gone,
too,</STRONG> there soon will be hardly any trace left of those
ill-conceived ideas of the past that Lynx could be used as a shell for
basic file manipulation etc. etc.</RANT>

Now to answer the burning question:

   AM I A DIRED MODE USER?
   =======================

Well, you probably are, if you are using Lynx on a unix system (it 
is not implemented for VMS) and DIRED has not been explicitly disabled -
the distributed top-level Makefile from the Lynx2-6 distribution has 
the following lines:

# To disable DirEd completely, prepend "#" to this line:
DIRED_SUPPORT     =   -DDIRED_SUPPORT

in other words, it is enabled by default.  (See that Makefile for more
information on DIRED and related configuration options, also, as usual,
see the User's Guide.)  Now, with DIRED enabled, you are in DIRED mode
whenever you use Lynx to view a local directory - whether you know it
or not.  In this mode, you have several additional functions at your
disposal, which you can invoke with the keys (by default) 'c', 'm',
'r', 'f', 'u', 't' and '.' - well approximately, the 'K'EYMAP page will
show you exactly which definitions are in effect.  Also, the 'e'dit 
function will act differently.  Again, see the User's Guide for what 
the functions are supposed to do.

Be aware that some of those keys are not working as they should if
you are using the default Lynx2.6 distribution.  You may even be able
to make Lynx exit with a Fatal Error by using the 't' or '.' key.
For some reported bugs, see 

        URL: http://lynx.cc.ukans.edu/lynx-dev/9610/0580.html

(and references therein).  For the fixes under discussion, see (currently)
Subir Grewal's page:

   Linkname: Patches and plugins for Lynx
        URL: http://www.nyu.edu/pages/wsn/subir/lynx/patches.html


Now, is someody gonna try those damn patches and report on it, or do I
have to go on and on in long-winded messages about an obscure feature
which doesn't really deserve such a lot of attention and which I was
hardly using myself?  :) 

   Klaus

P.S> Why I care, I don't really know, but I thought Martin Voges, who
did a good job about reporting those bugs, deserved an answer, rather
than silence.

;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;