13 March 2005
When I installed gforth-0.6.2 on Debian 3.0 Linux, I had some difficulty installing the info docs for gforth and vmgen. I found a simple remedy, but I do not know whether it was a good one.
During su -c 'make install', install-info aborted with an odd message about a failed directory lock during a file edit, and the Makefile aborted near the end. As far as I can tell, the install-info program of Debian 3.0r3 does not generate an empty info 'dir' file as needed. Also, the Debian /etc/profile did not define an INFOPATH environment variable to help info find the new docs.
I must study the install-info problem more. Meanwhile, I changed two things and got a good gforth install:
(1.) Before compiling gforth, I put an empty info 'dir' file in /usr/local/info/ by copying the system top level 'dir' file from /usr/info/ and deleting everything past the '* Menu' heading in the copy.
(2.) I defined and exported a global INFOPATH environment variable in /etc/profile.
The info system now finds the gforth and vmgen info docs if they are explicitly named (info gforth, etc), but gforth and vmgen do not appear in the top level info directory display.
With a suitably altered command in the gforth Makefile, install-info could put gforth and vmgen in the top level info directory file in /usr/info/. Another Makefile parameter to specify the location of the directory file would be needed. Maybe some Makefile parameters that specify the location of the gforth info docs would need to be renamed to clearly distinguish, on the one hand, the working location of the gforth info doc data and the operation of copying the data and, on the other hand, the location of the system info directory file and the operation of installing references to the new docs. Parameters named infotop and infodoc might serve, and the Makefile would need very little alteration, I think.
I feel a little uneasy about adding the global environment variable INFOPATH to the system configuration, because I have heard of secuity breaches that use altered parameter strings to produce an exploitable malfunction. Is this safe? For some reason, the builders of Debian 3.0r3 did not use it.
David Arnold
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
