[Patch, fortran] PR51976 - [F2003] Support deferred-length character components of derived types (allocatable string length)
Damian Rouson
rouson@stanford.edu
Tue Mar 4 06:05:00 GMT 2014
> Btw, there was quite some user feedback along the lines of "we want
> this feature", but there was not a lot of actual testing. So, if you
> are interested in this feature, please try the patch and report back
> here with things like "I tested it and it works great on my 100kLOC
> code" or "I tested it and found the following problem …"
The fundamental problem here is the pain in the !#@$#%!#$% it is to build gcc from source. Every time I try to do so, it costs me at least half a day and often more. For example, I just tried “svn update” inside my copy of the source and then followed written steps that worked for me to build from source just two months ago. As has happened repeatedly, I failed. Either something has changed in the set-up or (much more likely) my notes are missing some crucial element of what I did before. To be sure, this is my own failure, but I’m certain that I’m not the only person who has encountered this and most users who won’t last through the experience to get to the point of applying a patch and testing.
I’m not sure of the solution. I do very much wish gcc would move from autotools to CMake, but obviously I don’t expect to be able to influence that, especially given that I don’t contribute source, which loops me back to the previous paragraph: I’d be more likely to contribute as a developer if just building the source weren’t so painful.
Possibly a more reasonable solution is for someone to provide users accounts on a system that has candidate patches and to offer access to the system for users to test various patches before they are applied to the trunk. I imagine there are security concerns associated with offering accounts so liberally, but I imagine those concerns could be mitigated by putting the build inside a virtual machine to wall it off from the underlying actual machine.
I’ll be very glad to test the patch once it hits the trunk, at which point I’ll likely have access to it within a week via macports.
Damian
More information about the Fortran
mailing list