[net.unix-wizards] Help with a bug in nroff sought

cosell@bbn-labs-b.ARPA (05/11/84)

From:  Bernie Cosell <cosell@bbn-labs-b.ARPA>

I've tried this on our local nroff wizards and they don't know anything about
it - anyone out there ever seen this bug or maybe even fixed it?

Thanks
   /Bernie

----BEGINNING OF FORWARDED MESSAGES----
Date: Mon, 30 Apr 84 14:03:21 EDT
From: Bernie Cosell <cosell@bbn-labs-b>
Subject: bug explanation sought
To: nroff@bbn-labs-b

In trying to debug a bunch of macros, I was going nuts because I apparently
blundered into a bug in nroff.  Can anyone confirm just exactly what this bug
is and how it works?

Consider:
    .de X
    .so /etc/motd
    .tm it never gets here
    ..


    .X

When you place the call to macro X as the VERY LAST line in the input file, it
never returns from the .so and so DOESN'T get to the .tm -- if the call occurs
anywhere else in the source file, it works as you would expect and the .tm,
desipite what it says, does get done.  It seems clear that there is something a
little odd going on in its figuring out where to pick things up when you pop
out of .so - if the main source file is at EOF, that apparently prevents the
pop from being able to figure out where to resume the macro it was in....

Can anyone shed any light on this?

Thanks
   /Bernie

----END OF FORWARDED MESSAGES----

jeff@ism780.UUCP (05/16/84)

#R:sri-arpa:-65000:ism780:14400004:000:489
ism780!jeff    May 15 13:44:00 1984

>  From: Bernie Cosell <cosell@bbn-labs-b>
>  Subject: bug explanation sought
>  To: nroff@bbn-labs-b
>  In trying to debug a bunch of macros, I was going nuts because I apparently
>  blundered into a bug in nroff.  Can anyone confirm just exactly what this bug
>  is and how it works?
>      .de X
>      .so /etc/motd
>      .tm it never gets here
>      ..
>      .X

Is the input file >64k?  Are you running on a 16-bit machine?
The n/troff sources neglect to declare lseek() as long.

jeff@ism780.UUCP (05/17/84)

#R:sri-arpa:-65000:ism780:14400005:000:196
ism780!jeff    May 15 16:16:00 1984

>  Is the input file >64k?  Are you running on a 16-bit machine?
>  The n/troff sources neglect to declare lseek() as long.

    read ">32k" for "64k" in my original response to Bernie's message.