Re: Problem writing python tests without installing

[Doug Evans <dje at google dot com>]

On Thu, Nov 10, 2011 at 7:54 AM, Phil Muldoon <pmuldoon@redhat.com> wrote:
> Robert Lupton the Good <rhl@astro.princeton.edu> writes:
>
>> When running the testsuite, the python files that are sourced are apparently e.g.
>> ? ? ? gdb/testsuite/../data-directory/python/gdb/printing.py
>> rather than
>> ? ? ? gdb/python/lib/gdb/printing.py
>> although the binary is as expected gdb/gdb
>
> Right, but you don't have to do 'make install' to get data-directory.
>
>>
>> The files appear to be installed by "make install". ?This is another
>> manifestation of the problem that you cannot apparently work on the
>> gdb python code without first installing it (away from it's cvs home
>> -- but that can be faked with a link...)
>
> I work with Python every day and I don't have this issue. ?The files are
> correctly sourced from ../data-dictionary.
>
> I could not find the gdb.printing test in my output, but this works fine:
>
> python
>>Sending "import gdb.command.prompt" to gdb
> Looking to match ""[\r\n]*()[\r\n]+(\(gdb\) | *>)$""
> Message is "prompt substitution readline - import gdb.command.prompt"
> board_info host exists name
>
> FWIW GDB is spawned in the test as:
>
> Spawning /home/pmuldoon/bugs/setex/obj/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory /home/pmuldoon/bugs/setex/obj/gdb/testsuite/../data-directory
>
>> Is there a proper way to work around this?
>
> There's not enough information on your install/setup to diagnose the
> issue. How did you configure the build?

Yeah, I suspect the problem is either an absence of "-data-directory
..." when starting gdb, or not doing a "make" after editing a .py file
(*1).
[And if you add new .py files, remember to mention them in
data-directory/Makefile.in]

---
(*1) Since .py files aren't "compiled" having to do a "make" may be
unexpected, but it's ultimately easier to stage them in
$build/data-directory rather than trying to fetch them from the the
source tree.