Re: [Groff] CVS Build Problem

[Keith MARSHALL]

Zvezdan Petkovic wrote, quoting me:
> On Fri, May 06, 2005 at 03:51:17AM +0100, Keith Marshall wrote:
>> On Thursday 05 May 2005 7:40 pm, Zvezdan Petkovic wrote:
>>> BTW, I don't think Cygwin should be a measure of Unix compatibility.
>> 
>> Possibly not; but, like it or not, Microsoft Windows has the largest 
user 
>> base of any OS on the planet.  For those of us who are forced, out of 
>> corporate necessity rather than personal choice, to use it, Cygwin is 
>> probably the most widely used UNIX emulator, allowing us to use our 
preferred 
>> UNIXy tools on an alien platform.  You may not consider it a measure of 
UNIX 
>> compatibility, but you certainly can't ignore it as a measure of 
portability.
> 
> I am absolutely aware of that.
> Unfortunately, in terms of Unix compatibility Cygwin assumes too many
> things as if it's running on a Linux box.
> In those terms Microsoft SFU (Services for Unix) and UWIN from AT&T
> (David Korn) are much more Unix than Cygwin.
> That's what I meant.
> 
> I am also aware that much more people use Cygwin than SFU or UWIN,
> but we could also say that much more people uses Linux than Solaris,
> which doesn't mean that Linux is a better measure of POSIX conformance.

I think you may have misunderstood my point.  The goal is not merely to
ensure POSIX conformance, nor portability of groff to Linux, or to
Solaris, or to Cygwin, or indeed to *any* specific OS platform; our goal
is to make groff portable to as many platforms as possible.  I test on
Linux, Cygwin and MSYS, and to a more limited extent on SunOS, because
these are systems which I, personally, am able to test on.  I don't
claim that this guarantees any particular level of POSIX, or even UNIX,
conformance; what I *do* claim is that, in my experience, groff is
portable to these specific platforms, on which I am able to test.

> If you are going to ensure that groff runs on Cygwin, I do not see a
> reason why you wouldn't ensure that it runs under SFU.
> Exactly because of the corporate requirements, someone's boss may be
> much more comfortable with the (free) download of a software from
> Microsoft, supported by Microsoft, on the corporate machines.

Well, *my* boss has authorised me to install Cygwin and MSYS, but *not*
SFU.  I believe there is a perception that, since supporting UNIX is not
really in the interests of Microsoft's business plan, their commitment
to maintaining SFU may leave something to be desired.

> I'm using groff for Windows on my office machine, and I have a VMware
> with a whole virtual network of virtual Linux and BSD machines, so I
> didn't feel a need to install Cygwin.  Other machine in my office is an
> HP-UX workstation.  At home I run only Linux and OpenBSD.

Great.  So, in addition to confirming groff's suitability for use on
Linux, (which is pretty much a "given" in any case), you could also
confirm its portability to OpenBSD, and possibly HP-UX too.  As
developers, we tend to assume things work as expected, until someone
complains to the contrary; if no one reports the problems, then they
don't get fixed.

> However, if there's nobody else to do it, I wouldn't mind installing SFU
> on my office machine and from time to time compiling groff under it.

That too would be great.  AFAIK, no one has yet stepped up to verify
groff's portability to SFU, but neither has anyone reported problems
with it.  That could be because it works perfectly, or simply because
no one has tried it.

Best regards,
Keith.