05-02-2022, 12:23 AM
I think it's a very good suggestion and something I'd like to see fixed for the Linux version - We should probably replace
1. The ./internal/temp location is in more than a few different locations in the source, so those would need to be rounded up and addressed.
2. Some files outside of
3. Many object files for libqb are compiled outside of
I would definitely encourage you to submit an issue to the Github here, that would be the best way to ensure this idea doesn't get lost.
Also, as a workaround you might be able to turn the
./internal/tempwith respecting the TMPDIR environment variable and fallback to
/tmpitself, potentially with a command line option to manually specify the location. Unfortunately I think there are a decent number of barriers to this though:
1. The ./internal/temp location is in more than a few different locations in the source, so those would need to be rounded up and addressed.
2. Some files outside of
./internal/tempare modified during compilation, in particular qbx.cpp. qbx.cpp also has direct statements that include files from
./internal/temp, so really a full solution for dealing with this file is needed. There might also be one or two other stray files modified during compilation that also need addressed.
3. Many object files for libqb are compiled outside of
./internal/temp. These don't change as much as the stuff in
./internal/temp, but they might be annoying. If we offer a 'proper' Linux install those files could potentially be shipped precompiled so we can probably ignore them for the moment.
I would definitely encourage you to submit an issue to the Github here, that would be the best way to ensure this idea doesn't get lost.
Also, as a workaround you might be able to turn the
./internal/tempdirectory into a symlink to a location on your ramdisk. Not an ideal solution, but I think it would get you most of the way there without needing to wait for changes to QB64.