| Re: IDL & Win95 -- arrggh [message #6569 is a reply to message #6512] |
Fri, 12 July 1996 00:00   |
thompson
Messages: 584
Registered: August 1991
|
Senior Member |
|
|
Peter Mason <peterm@demsyd.syd.dem.csiro.au> writes:
> <snip>
>>> How about long procedure names? I would think this would be even worse
>>> than the long data file names. I have a huge idl application I've been
>>> working on for almost 3 years under various flavors of unix and on the
>>> Mac (> 12,000 lines of code). Lots of procedure names like
>>> "define_colors.pro", "set_contours.pro", "widget_window.pro," etc that
>>> don't conform to DOS limits.
>>
>> I haven't tried to use IDL under the newer flavors of Windows, but in Windows 3
>> one could reference a procedure by its name, and it would find it with the 8.3
>> filename. E.g., calling set_contours would look for a file called
>> "set_cont.pro". This causes problems when one has multiple routines which
>> start with the same eight characters.
> <snip>
> Sadly, this is bust under NT and 95 if you retain the source files' long
> names (on disk). But if you go and rename the files 8.3 style it'll work
> as under win3.1x.
> But things are not as bad as they seem - you don't have to rename all
> your source files to port to Windows if you're willing to "compile" your
> source and distribute a .SAV file instead of source. A .SAV file compiled on
> one platform will work on another, as long as the IDL versions are the same.
> If you develop under Unix and you use long names for your routines this
> won't be a problem under Windows - IDl calls a routine using the full
> routine name (once compiled).
> And the only filename you have to worry about is that of the .SAV file itself.
That would require a nightly cron job to create either many thousands of save
files for each procedure (and then how would we name those files?), or one huge
save file containing all the software, which would mean that people would have
reinstall everything if even only one routine changed. Either solution is the
kind of hack that we shouldn't have to go through. We also can't guarantee
that those who are trying to run our software would have exactly the same
version of IDL that we do.
Another disadvantage of such save files is that the routines we have for
exploring the documentation headers of the routines would not work.
Bill Thompson
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
