Message-Id: <199802141800.KAA08409@mailhost2.cac.washington.edu>
From: "Ned Ulbricht" <nedu AT ee DOT washington DOT edu>
Organization: University of Washington
To: Nate Eldredge <eldredge AT ap DOT net>, Eli Zatetskii <eliz AT is DOT elta DOT co DOT il>,
        George Foot <george DOT foot AT merton DOT oxford DOT ac DOT uk>
Date: Sat, 14 Feb 1998 10:00:18 -0800
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: Suggestion: Portability section for libc docs
CC: djgpp-workers AT delorie DOT com
References: <199802130328 DOT TAA12320 AT adit DOT ap DOT net>
In-reply-to: <Pine.OSF.3.95.980213034858.31350A-100000@sable.ox.ac.uk>
Precedence: bulk

On 13 Feb 98, George Foot wrote:

> On Thu, 12 Feb 1998, Nate Eldredge wrote:
> 
> > Also, maybe the header used by a DOS compiler should be in-line with the
> > column entry, instead of a footnote. I assume it will be a common occurence.
> 
> Provided the same header is used by all DOS compilers; if not a footnote
> would be more appropriate (and probably necessary anyway).
> 
> > Here's my revision of Eli's suggestion:
> > 
> >   @subheading Portability
> > 
> >   @multitable {Supported}     {ANSI}   {POSIX}   {Unix}   {MS-DOS/MS-Windows}
> >   @item                  @tab  ANSI @tab POSIX @tab Unix @tab MS-DOS/MS-Windows
> >   @item	      Supported? @tab   no  @tab yes @tab yes (1) @tab yes <io.h> (2)
> >   @end multitable
> [snip]
> 
> I'm wondering whether the `Supported?' bit is necessary, since the table
> only has one (real) row and it should be pretty obvious what `yes' and
> `no' mean in each column.  Also I think it's wise to keep tables fairly
> narrow when they're being converted into a markup language, since you
> don't know exactly what they'll look like (margins, page width, etc) on
> the user's screen.  This could be particularly relevant for RHIDE users; I
> don't use it myself but I presume its help window is resizable. 
[snip]
> @portability
> @brief ansi(no) posix(yes) unix(yes,1) dos(yes,io.h,2)
> @notes
> 
>   (1) SysV flavor doesn't frobnicate the foobar.  BSD does.  Many Unix
>       systems don't have the prototype declared anywhere (so it's best
>       to have an explicit prototype in the program).
> 
>   (2) Known to be buggy in Borland.
> 
> @end portability

Eli suggests MS-DOS/MS-Windows while George suggests just DOS for 
column heads.  Note that one of the differences is whether the conio 
functions are supported (they aren't for native Win3.1 programs 
although a DOS box can use them).  I would suggest that we ignore 
native Windows compatability since anyone using DJGPP for Win 
apps needs to have additional Win docs anyhow.

If it weren't for trying to keep the table narrow, I would actually 
suggest different columns for MS, Borland and Watcom.  If all three 
were 'yes' then it would be pretty much ms-dos compatible.  We'll 
want to keep that info somewhere anyhow, so that we know what's been 
compared to which docs.  (I'm shutting up about testing).  That is 
unless we just go with Eli's suggestion that anything MS has is by 
definition MS-DOS compatible--but then how would we know that Borland 
uses a screwy header file for instance?

Also, I assume that Unix means BSD--are there any differences among 
major Unix flavors that will actually make a difference?

BTW  Nate, Eli & George: I'm now subscribed to the djgpp-workers 
list so there's no longer any need to keep copying me 
explicitly--thanks for doing that guys!  I'm still copying you three 
in addition to the list--let me know if you want to me to stop.  (You 
can continue to copy me if you want, I just get rid of the 
duplicates).