🚀 go-pugleaf

Sorry it has taken so long to reply.

> I tried writing a file when the disk was full and it got an error,
> as it should.  19.19 is almost a year old; it's not worth investigating
> any further unless there's evidence of a problem in a recent version.

I just finished building version 19.22 on my Sun IPX running SunOS 4.1.3.  I  
tried building version 19.25 on the Sun, but it kept killing the Emacs process  
whenever I did a (suspend-emacs) or M-x suspend-emacs (I used gcc (2.6.0) --  
with GNU as/ld and without GNU as/ld -- and Sun's cc, but none of these fixed  
the problem and I didn't see it listed in the PROBLEMS file so I went  
backwards).  However, back to the topic at hand.  Version 19.22 (on the Sun,  
anyway) did detect an IO error while writing the file and did not exit.  So,  
this problem appears to be a non-issue now.  Thanks for looking at it.  I'll  
verify that it behaves the same way on SGIs as soon as I can get it (or 19.25)  
installed.

> If your company makes a lot of use of GNU software, would you think a
> donation to the FSF might be appropriate?
>
> An easy way to make a small contribution is to order distribution tapes
> or CD-ROMs from the FSF, or subscribe to periodic updates.

I would love to see this happen, but I don't expect management to see things  
the way I do especially since our company is trying to keep people covered  
(times are tough for aerospace companies).  I'm just a grunt in the machine.   
If were up to me, I'd be donating programmers and software.  I *will*,  
however, bring it to their attention (especially mentioning how fast you  
responded to my original message, which is much faster than any support we've  
ever paid for).  If I can be of service, please let me know.

Since we are talking about "problems", here is one that just bit me using GCC  
(which I'm cross mailing to the GCC problem account) and SGI's cc (who are  
also going to receive a copy):

---------------->8 Cut Here 8< ----------------
const int size = 4;

struct my_struct
{
  float foo[size];
  int   bar[size];
} my_struct;

int main(void)
{
  return 0;
}
---------------->8 Cut Here 8< ----------------
which generates the following errors (gcc 2.6.0):
[mrg@chartreuse]~> gcc t.c
t.c:5: variable-size type declared outside of any function
t.c:6: variable-size type declared outside of any function

I don't have an ANSI C manual handy, but I expected that a 'const int'  
wouldn't be considered a varying entity.  Of course, this works fine if you  
#define size 4, but then you lose symbol table information which is helpful  
during debugging (especially of someone else's code).  Maybe this is perfectly  
legal in ANSI C, but it sure would be nice if the constant were treated as a  
constant.  Maybe this is more an annoyance than a bug.

Thanks once again.
---
Michael Gentry               | Software Grunt        | Caffeine Addict
Coleman Research Corporation | 6820 Moquin Drive     | Huntsville, AL 35806
E-Mail: mrg@loki.hsv.crc.com | Voice: (205) 922-6000 | NeXTmail welcome
	"Did dinosaurs only use 10% of their pea-sized brain?" - mrg