Re: lynx-dev Win32 compilers

[Foteos Macrides]

David Combs <address@hidden> wrote:
>> didn't perceive my questions as criticisms.  Perhaps if I have further
>> questions you and Doug would be kind enough to answer direct email from
>> me rather than via lynx-dev.
>> 
>>                                         Fote
> 
>No, please;  Fote, the stuff you subimit is good and educational
>for all of us, including questions about msdos.  We all learn
>from it; if not it is quick enough to just go to next email.
> 
>Yes, there has been some nasty stuff passing through this list,
>but is again quick enough to go to next email.   Likewise with
>all the spam with stealth subject-lines.
 
        Well, OK.  As far as the compiler issue is concerned,
Lynx historically has been developed with the aim of not
requiring a particular compiler on any platform.  On VMS,
all three of the compilers available to it (DECC, VAXC, and
GNUC) have been suppported (I don't know if that's still
so in v2.8), and the autoconf in v2.8 supposedly makes Lynx
compiler-independent on Unix.  On MSDOS (with or without a
Windows API) my impression is that Borland historically has
been the compiler of choice by developers (e.g., Craig used
that for developing DosLynx) and so it would be surprising
if Wayne had not used that for Bobcat, and then similarly
for his contributions in the devel code.  For straight DOS,
djgpp appears favored in the present absence of a WINSOCK
equivalent.  The cygnu development is recent, and it's home
page still leads you to a long list of "known problems" with
it, but apparently it's now stable enough to be used seriously,
and further enhanced by the MINGW32 interface.  I presume
MicroSoft's Visual C/C++ also would be workable in conjunction
with WINSOCK.  I think it would be in the best long-term
interest of Lynx if the objective were to make it as compiler
independent as possibile for DOS/WIN/NT, but I don't think
it's reasonable to expect that Wayne, Doug and Bill undertake
that objective by themselves.

        When all is said done, the autoconf in v2.8 is only
one-third done (it doesn't yet extend adequated to the
DOS/WIN/NT ports, and not at all to VMS), and I personally
think that's more due to an underlying Unix-centrism than to
lack of time and resources.  That conflicts with the original
objective of the Web (platform-independent information sharing)
and Lynx's historical development (make it easy for anyone to
contribute, no matter what compiler is installed on the
contributor's system).  But again, I don't think it's reasonable
to expect that Wayne, Doug and Bill take that on for DOS/WIN/NT
by themselves (and I don't think Brian has any chance of inducing
Richard Levitte to extend the autoconf to VMS if TD, instead of
Richard, keeps replying that he doesn't have the "time"; Richard
has long lurked on this list :).

                                Fote
--
Foteos Macrides (address@hidden during April, '98)