[Info-vax] CLI editing, was: Re: VMS - Virtual Terminals - A security risk way back yonder OR was that an Old Wives Tale ?

[John E. Malmberg]

On 2/15/2016 9:00 AM, William Pechter wrote:
>
> Which is why I liked the old rn/elm configure scripts from the pre-gnu
> configure 80's which would recognise Eunice running under VMS.
>
> Gnu configure is only useful on more Unix-y like systems of the 90's.

Gnu configure is not performing the minimum optimal tests to accomplish 
what the program needs.  For some reason it is trying to test many 
issues that either do not matter, or that it already knows the answer to.

Recent GNU configure scripts first test to see if a header file is 
usable and regardless of the outcome, perform essentially the same test 
to determine if the header file exists.

GNU configure is also testing to see if the system library calls work 
with out the header files being used before testing to see if the system 
header files are present and working for the libraries.

As a result, GNU configure will generally not produce the optimal 
configuration for any system with backwards binary compatibility.

Since on *nix systems, the resulting configuration will usually work, it 
may not get noticed that it is not the best.

On VMS, this is usually noticed because the alternate routines that 
configure selects are usually not present.

It is for this reason running GNU configure scripts on VMS require 
helper files that are looked for by the ld_tools ld/cc wrapper.

At a minimum a conftest.c_first file is needed to make sure that the 
posix variant of the exit() routine is called, or otherwise many of the 
tests in configure will return the wrong answer.

Regards,
-John
wb8tyw at qsl.network