LYNX-DEV modular code

[Klaus Weide]

On Sat, 9 Nov 1996, Nelson Henry Eric wrote:

> I've been trying to hang low, but as the pseudo-quasi-sysadm of a
> limited-resource machine I'd like to make a humble request to the
> lynx-dev community.  The multi-bookmarks feature is very convenient
> and probably should be in the standard lynx, but it may not be useful
> to everyone.
> 
> Refer to Klaus's post for particulars:
>       http://129.237.17.75/lynx-dev/9610/0669.html.
> What I would like to see is a more modular structure of lynx code,
> with an easy way of not including modules that are unnecessary or
> even undesirable for a particular site.
> 
> It is extremely easy to remove modules from the wwwlib stuff; there are
> even commented instructions right along the code in HTAccess.c on how to
> do it.  Of course this has no effect on the lynx binary.  

Do you mean no effect at all, or just no significant effect?  If you
eliminate e.g. all traces of HTFinger.o from the relevant Makefile,
there is no way that module can get included in the Lynx binary, so
it should definitly become smaller.

> I've been trying
> to take out, for example, the LYTraversal.o module, which _would_ have a
> significant impact on the final size of the lynx binary.  However this
> effort has proven to be no picnic, and quite frankly I still haven't
> succeeded.

Without specifics on what you tried, and how it failed, there's no way
of helping you..  (well unless someone would try the same thing
independently, and run into the same problems).

> So what I am asking is, could new major features be appended to lynx as
> modules with a simple mechanism for not including them?  (Or is that
> already being done, but I don't know how to do it?)  The only concrete
> example I could give would be Rob's styles enhancement, which increases the
> size of the lynx here by over 10%.  Does just dropping the -DCS from the
> compile flags give the same effect as, in addition, removing `($STYLE).o'
> from src/Makefile?  (I've tried to test it, but I cannot get lynx-rp to
> compile.)  Could macros be set up to remove all unwanted modules, with the
> added feature of providing messages "%% is not supported" should someone
> try to use a feature that is available in the standard lynx?

But I thing Rob is trying to do just that with his additions - making
it possible to exclude what you don't want by using lots of #ifdef's.
If that doesn't work for you, I guess you have to figure that out with
him.

Doing this at the time when a new feature is added is relatively easy.
Trying to do it later, as an afterthought, is more difficult.

> What puzzles me as a layman is, why couldn't _all_ of the unwanted code,
> eg ftp, telnet, mail, news etc. etc. when running -validate, be removed
> from the code in the first place rather than blocking its use?

That sounds like you want dynamic libraries.  And then I don't even
know whether they could do what you want..  At least it would be
more system dependent than the current "everything in one binary"
procedure.  And it would probably need lots of code modifications,
shifting declarations and refenrences around among files..  

  Klaus  

  

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