Fix 300: delete the contents of $tmpdir after creating it

[realtime-neil]

No description provided.

[realtime-neil]

@megastep the suidtest is failing because it apparently depends on the target directory not being emptied prior to extraction?

Can you help me understand the use case that requires the preservation of files (suid or otherwise) within the target directory?

[megastep]

While I didn't write this test myself, it appears it is failing because the file disappearing would make the comparison of the permissions before and after extraction fail. Obviously a removed file no longer has permissions...

[megastep]

I think this was meant to test that Makeself wouldn't overwrite permissions of any existing files after extraction. So in that sense obliterating the whole directory would break this. If we explicitly allow for this possibility when using certain command line flags maybe that test should be revised.

[megastep]

Maybe a simple solution would be to have the suidtest use the --notemp argument.

[realtime-neil]

Maybe a simple solution would be to have the suidtest use the --notemp argument.

Can do. On it...

[realtime-neil]

@megastep , I think #300 and suidtest are fundamentally incompatible.

The suidtest places a file within the extraction directory and asserts a very specific mode bit (the SUID bit) thereon is preserved during extraction.

In issue #300, @belonesox proposes the non-temporary extraction directory should be emptied just prior to extraction.

I can admit my bias is with the status quo in this conflict: the simplest thing to do during extraction is nothing at all and doing nothing at all means not cleaning the extraction directory prior to extraction.