ECPG regression tests need .gitignore update?

Robert Haas <robertmhaas@gmail.com> writes:

After running "make installcheck-world", git status shows a bunch of
stuff that looks like this:

# src/interfaces/ecpg/test/sql/desc.dSYM/
# src/interfaces/ecpg/test/sql/describe.dSYM/
# src/interfaces/ecpg/test/sql/dynalloc.dSYM/
# src/interfaces/ecpg/test/sql/dynalloc2.dSYM/
# src/interfaces/ecpg/test/sql/dyntest.dSYM/

Mac user eh?

This seems like a build-process bug, not a .gitignore oversight.
Those shouldn't be there at all (and I notice "make clean" doesn't
get rid of them).

regards, tom lane

On Tue, Oct 19, 2010 at 11:27 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:

Robert Haas <robertmhaas@gmail.com> writes:

After running "make installcheck-world", git status shows a bunch of
stuff that looks like this:

#   src/interfaces/ecpg/test/sql/desc.dSYM/
#   src/interfaces/ecpg/test/sql/describe.dSYM/
#   src/interfaces/ecpg/test/sql/dynalloc.dSYM/
#   src/interfaces/ecpg/test/sql/dynalloc2.dSYM/
#   src/interfaces/ecpg/test/sql/dyntest.dSYM/

Mac user eh?

Blame dpage.

This seems like a build-process bug, not a .gitignore oversight.
Those shouldn't be there at all (and I notice "make clean" doesn't
get rid of them).

I don't even know what they are.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Robert Haas <robertmhaas@gmail.com> writes:

On Tue, Oct 19, 2010 at 11:27 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:

This seems like a build-process bug, not a .gitignore oversight.
Those shouldn't be there at all (and I notice "make clean" doesn't
get rid of them).

I don't even know what they are.

"Outboard" debug symbols.

Some googling just turned up this interesting comment:

Interesting quirk: on MacOS 10.6.2, if you include a source file in the
'link' line, then the dsymutil program is run to generate the .dSYM
information; if you only link object files and libraries, then the
dsymutil program is not run so the .dSYM information is not
generated. You can validate this with the '-v' option to GCC which shows
the programs executed by 'gcc'.

So maybe we can get rid of it by not using the "shortcut" of skipping
creation of a .o file. Off to try that ...

regards, tom lane