Re: LYNX-DEV BSD makefile support 3

[Bela Lubkin]

Michael Sokolov wrote:

>    I know that new BSDs support both East Coast and West Coast semantics. I
...
> ago), but to help other devotees of the West Coast semantics.
...
>    When both semantics are available, the real issue is which one is
> better. However, I'm not trying to start a flame war proving that BSD is
> better than POSIX, although I do believe that. I want users do decide for
> themselves. Since both the existing target and the one I'm proposing work
> fine on BSD systems, the real issue is which one is better.

I have approximately zero interest in a divisive argument about which
"coast" sponsors which syntax in one small utility in an operating
system.  The configure script figures it out.  It has no political
leanings.

>    A note on BSD systems. You claim that only BSDI doesn't use the GNU
> semantics. However, you have forgotten the most important BSD system of
> all: the pure official 4.4BSD tape from UC Berkeley. Although I don't have
> it now, that's what I'll move to in the near future when I get my new hard
> disk drive. That system contains very few GNU/POSIX impurities.

What you label "impurities" are additions which make the software
usable.  Perhaps you would also like to move back to Lynx version 2.3.6
(or so), which has no "GPL impurities".

>    I know that Lynx is maintained by East Coast devotees, and I can imagine

Uh-uh.  The only label you can usefully apply to the maintainers of Lynx
is "Lynx devotees".  Anything else is baggage imposed by your own
colored lens of the world.

> that you think that POSIX/GNU semantics are better and should be used
> whenever possible. However, think of it this way. This issue concerns the
> people who compile Lynx, not those who just use it. These people are either
> sysadmins or people with their own UNIX boxes. In both cases, these people
> make the choice of which operating system to install and use.
>    The only way a person can use the 4.4BSD target is if he/she has made a
> deliberate choice against Linux and even FreeBSD/NetBSD in favor of the
> pure 4.4BSD tape from UC Berkeley. This means that he/she believes in the
> West Coast way of doing things strongly enough to go to the trouble of
> getting an encumbered tape and solving installation puzzles (pure 4.4BSD is
> VERY hard to install) instead of getting FreeBSD or NetBSD for free and
> installing it painlessly. This means that such a person is a real BSD die-
> hard and probably detests GNU and POSIX. I'm sure that such a person would
> prefer Lynx to compile without -DSVR4 and to use the BSD makefile include
> semantics.

I'm sure that all two of you will be very happy when that is done.

>    I claim that such a person is the only kind that will need the 4.4BSD
> target. If a person doesn't belong to this extreme category, he/she will
> install Linux or at least FreeBSD/NetBSD and won't need the 4.4BSD target.
> Therefore, adding the target I propose won't hurt anyone but will help the
> people described in the previous paragraph.
>    Also, it will help BSDI users, since my target reflects the true nature
> of BSDI more accurately than the existing one.

As far as I can tell, you want to introduce changes which have no
material effect on the compilation or execution of Lynx, merely to
preserve the ideological purity of the process by which it is built.
Nobody "needs" these changes.

I have a different ideological reason to want to eliminate "-DSVR4"; and
my ideology appears to be in alignment with that of the people
(T.E.Dickey and Jim Spath) who are working on autoconfig for Lynx.  My
ideological point is that the concept "SVR4" does not in any useful way
capture the behavioral changes introduced by the symbol.  It is an
essentially arbitrary string that was attached to those behaviors by the
person who first introduced them.  Those attached behaviors in fact
apply to all sorts of other POSIX-like systems, including the BSD
systems you are so vigorously defending.  (BTW, apparently the 4.4BSD
developers didn't even agree with your ideology: one of the very goals
of the 4.4 development cycle was to incorporate POSIX semantics.)

Anyway, to return to my point: "SVR4" is a bad tag.  In the current
source, it encapsulates three behavioral changes at compile time:

  #define USE_DIRENT
  #include <sys/fcntl.h>
  #include <time.h>

That's it.  Unless those three directives are wrong for BSD -- and I
don't think that's the case -- the tag "SVR4" isn't causing any problem.
The tag is slowly being abolished by the autoconf work.  For now, just
ignore it.

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