home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
TeX 1995 July
/
TeX CD-ROM July 1995 (Disc 1)(Walnut Creek)(1995).ISO
/
tex-k
/
tex-k-archive.past
/
tex-k-archive.gz
/
tex-k-archive
/
000811_kb@cs.umb.edu_Fri Jul 29 02:47:06 1994.msg
< prev
next >
Wrap
Internet Message Format
|
1994-10-11
|
3KB
Received: from terminus.cs.umb.edu by cs.umb.edu with SMTP id AA21403
(5.65c/IDA-1.4.4 for <tex-k-exp@cs.umb.edu>); Fri, 29 Jul 1994 06:47:07 -0400
Received: by terminus.cs.umb.edu id AA15575
(5.65c/IDA-1.4.4 for tex-k); Fri, 29 Jul 1994 06:47:06 -0400
Date: Fri, 29 Jul 1994 06:47:06 -0400
From: "K. Berry" <kb@cs.umb.edu>
Message-Id: <199407291047.AA15575@terminus.cs.umb.edu>
To: tex-k@cs.umb.edu
Subject: Re: \font-bug in UNIX-TeX
Sorry I wasn't able to reply soon on this.
\font\xxx=cmr10{\xxx test}\bye
The problem is, that TeX wants to read the metric information
of a file `cmr10{'
As Pierre says, filenames in TeX are terminated by spaces. Though this
is supposedly a system-dependent area, in practice it would cause
enormous difficulty to try to recognize the end of a filename in any
other way.
Anyway, to get to the real question:
2. But why then passing the ${abc} through shell-expansion, while
backquotes not run the command.
It's not run through a shell via system(3), Pierre is right about
that. But what I have snuck past him is that the filenames are run
through kpathsea, which understands ~, ~foo, $foo, and ${foo} like a
shell, but nothing else.
I think this is a win. I hope everyone agrees :-)
The last statement holds
for Solaris 2.3. On an RS6000 with AIX 3.2.5 I got the file
`b${abc}' read. So it seems to be a feature of Solaris.
Say what?! The same version of web2c? If the two tex's use the same
kpathsea, then they should act in the same way. If not, I am totally at
a loss.
Why does the brace
not have the catcode 1 in filenames
Because that's the way Knuth implemented the \input primitive. Sorry.
There is no way to change it, short of
\let\primitiveinput=\input
\def\input#1{\primitiveinput #1 \ignorespaces}
... and even that isn't precisely equivalent to the weird semantics of \input.
This shows that in filenames the blank is always a delimiter,
not depending on its catcode.
Ooh, yuck. Guess DEK didn't bother to program that case for \input.
So I see not how could read
arbitrary filenames
Since your catcode test didn't work, I don't think you can. The space
will always delimit it. I can't think of any feasible way to change this.
There have been occasional calls
for a "system" interface through TeX. I doubt
that it is worth the pain it would involve.
It's easy to do in its most trivial form -- \write18{foo} does
system(foo). I think this is the first extension planned for e-tex,
because it is so easy to implement.