This is the mail archive of the binutils@sources.redhat.com mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: RFA: ARM dynamic object tests


On Sat, 3 Jan 2004, Daniel Jacobowitz wrote:
> By the way, ld-cris.exp is technically wrong. The use of exec means you
> != can't run the tests with build host; instead you have to run the
> tests on the host.  Hmm, I see that the rest of the ld tests are
> absolutely littered with this assumption too, so I guess it really
> makes no difference.

Note the use of exec in run_ld_link_tests.  And run_dump_test. ;-)

Hum, Canadian cross ld testing isn't supposed to happen on the
build host, is it?  Anyway, Canadian cross testing... *shrug*.
(Meaning: I'd rather lose it completely than letting it stand in
the way of testsuite maintenance.)

> I don't have any problems reading the input of run_ld_link_tests,

Do you mean the option lists (in the test-caller file) or do you
mean something else?

> which
> you'll notice is in lib/ld-lib.exp anyway and therefore just as much a
> first-class citizen as run_dump_tests.

Sure, it is.  It's just that run_ld_link_tests is the kind of
hack from which run_dump_test evolved (really: adjusting it from
the gas testsuite and including it in the ld testsuite) and IMHO
improves maintenance for *most* tests; notably those not
re-using previous test results.

>  It offers the ability to save
> the output file which was all I needed.

I guess we can add the ability of naming output to
run_dump_test, or other support for re-use of linked files.  Or
add the ability of reading options from input files to
run_ld_link_tests.  :-)

Anyway, I just wanted you to consider run_dump_test, which you
did, but didn't mention it at first AFAIR.

brgds, H-P


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]