I'm a windows user running gfortran. I've been creating and updating
for several years now for use on windows under gfortran. Very nice BUT the task
has been made somewhat
difficult because the makefiles dont work. They are also somewhat opaque in
style ("techy"), especially to
windows users who are not familiar with makefiles or unix.
With some effort it is possible to create the libraries, even relatively
straightforward, only a few tricks need unraveling.
But the testing routines are quite a tangle, and basically the makefiles must
be run (but they choke).
It was possible to untangle things in 3.2 (2 command line changes in two files)
and publish the fixes for
windows folks, and in 3.2.1 a few more problems arose and a complete review of
everything was required,
so that batchfile used to replace what could not be repaired. The
incompatibilities seem to again be escalating
in 3.2.2 and I'm wondering if these incompatibility problems are really
necessary, or more an artifact of a new
person working on the makefiles.
A particular problem, but not the sole problem, is that the programmer
modifying the make-files appears to be
using unix commands such as 'mv'. Only unix systems with the mv utility can use
these makefiles! Certainly
windows has copy and rename functions, but they are not called 'mv' nor
function with a necessarily compatible
syntax. While it is true that what is necessary must be done, alas, it all
seems unnecessary, for my review suggests
that the use of mv is actually gratuitous, that is, alternatives could be used.
There also seem to be big swings in the
style and logic in parts which may be cool from a unix perspective, but are
confusing and opaque for someone trying to
follow the logic.
Perhaps the programmer would be willing to contact me to help identify places
where there are problems and we
could work toward sorting some of these out and helping make the package more
compatible and usable to windows
-------------- next part --------------
An HTML attachment was scrubbed...