Re: Patch for pascal-dynamic arrays

[Jan Kratochvil <jan.kratochvil@redhat.com>]

On Wed, 16 Sep 2009 20:17:38 +0200, Joost van der Sluis wrote:
> I had tested everything without optimisation (for debugging) and -Werror
> because that doesn't compile on windows.

Could you please run it also on GNU/Linux as some final test?  If you do not
have one available there are various test boxes available at sourceforge.net
and similar sites or contact me offlist.  Testsuite on GNU/Linux should IMO
catch most of the possible regressions.


> > Then also please check regressions of the testsuite - `make -C gdb check' and
> > comparing gdb.sum before/after the patch.
> 
> About the testsuite... it takes ages to complete.

One `make check' should take at most about 20 minutes in worst cases, some
5 minutes should be more normal.  Sure it is being run just as a final test.
Many FAIL results are normal, just regressions are worth a check.


> Is there some way to run just one test (or just one .exp test-file?)

I use
	cd gdb/testsuite
	make site.exp	# needed just once
	runtest runtest gdb.base/start.exp
but probably more correct is to just run:
	cd gdb/testsuite
	make check RUNTESTFLAGS="gdb.base/start.exp"


> I don't know anything about the GNU coding style,

No problem just please try to follow the spacing/indentation of the code
around despite it looks unusual.
	http://www.gnu.org/prep/standards/standards.html#Formatting


> But I think that there are only a few issues which will fix a lot of
> regressions. I'll have a look at them.

Yes, I agree, it may be even a single problem.


> Just a remark: some of your changes in the coding-style are corrections
> for portions of the code that I copy-pasted from other parts.

:-)  Some existing GDB code may not follow it but I do not know much of such
places offhand.


(The GNU coding style really was not a primary concern of my mail, feel free
to keep the code parts which you are not sure about the formatting intact.)


Thanks,
Jan