bug#3700: 23.1.50; verbose execution of configure script, determining the maximal length of command arguments

[Peter Dyballa]

Hello!

The configure script of (lib)m17n-lib 1.5.4 is buggy and fails at  
line #32967. The report was unclear or misleading (I found that the  
function PKG_CHECK_MODULES() does not seem to be defined), so I tried  
to run it in a *compilation* buffer as 'sh -x ./configure' – and  
after twelve hours it hasn't finished. Within minutes it finished in  
GNU Emacs 21.3.50 and 22.3 – and also in *shell* buffer in GNU Emacs  
23.1.50. In the *compilation* buffers of GNU Emacs 23.0.60, 23.0.94,  
and 23.1.50 the progress is slower than a snail when the script tries  
in a loop to determine the maximal length of command arguments.  
"Words" with a length of a few K obviously cause this stop.

For the test I launched all five Emacsen with -Q, opened the prepared  
copy of the original directory in dired-mode, invoked M-x compile RET  
and changed the default 'make -k' into 'time nice +19 sh -x ./ 
configure' without pressing RET at once. After all lines were  
prepared in mini-buffer I pressed RET in each.

The GNU Emacsen 23.x are all not really responsive (the system is, I  
can move around the frames, I can resize them, but the windows do not  
refresh), changing to the other window works, but reaching the end of  
the buffer might succeed tomorrow. When GNU Emacs 23.1.50 was running  
with all my customisation, the time in mode-line was updated in  
intervals of around 10 min.

GNU Emacs 23.0.60 refreshes fastest…

--
Greetings
  Pete

Without vi there is only GNU Emacs