Re: [Vala] little architecture bug in valac

[Raffaele Sandrini <raffaele sandrini ch>]

Welcome on the vala list.

On Mon, 2007-08-13 at 13:34 +0200, Piotr Skamruk wrote:
> it's my first post to this list so at start i should say: hi ;)
>
> look at this little session with valac:
>   jell jells:~/Dane/src/gnome/vala-tests$ valac -C --pkg=gtk+-2.0 sample.vala 
>   Compilation succeeded - 0 warning(s)
>   jell jells:~/Dane/src/gnome/vala-tests$ ls
>   sample.c  sample.h  sample.vala
>   jell jells:~/Dane/src/gnome/vala-tests$ valac --pkg=gtk+-2.0 sample.vala -o sample
>   Compilation succeeded - 0 warning(s)
>   jell jells:~/Dane/src/gnome/vala-tests$ ls
>   sample  sample.vala
>
> after second valac invocation files generated firstly was removed, but
> correct way should be to not even touch this files.
>
> i think that .c, .h, .o files should be generated in temporary directory
> (perferably in /tmp/, with name constructed by mkdtemp invocation), and
> only after succesfully generated output file, it and _only_ it should be
> moved to destination directory. of course in some finalization stage
> temporary directory should be removed.
We are aware of that. This has to be fixed. Its not a behavioral
decision to take over those files.

> btw. another unusual manner of valac is that it informs user on stderr
> that there was no error in time of compilation. i think that after
> succesful compilation onest information should be exit status equal to 0
> unless there would be some verbosity levels increased by user in
> invocation time...
This would be more in the GCC manner.
The vala compiler tries to act similar to the mono C# compiler in terms
of console output.

Raffaele