feature request: make check in parallel about flint2 HOT 5 CLOSED

On 18 March 2014 23:10, jakobkroeker [email protected] wrote:

it seems that Flint builds in parallel using e.g. make -j4,
is that safe?

Yes.

definitely not working perfectly is make -j4 check, since the output gets
scrambled.
Is it possible to fix that?

Why? It's not broken.

Maybe 'GNU parallel' could be useful, but I'm not sure about this.

Reply to this email directly or view it on GitHubhttps://github.com//issues/54
.

from flint2.

Why? It's not broken.

doh, probably you are right
the output gets scrambled (see below) but if a test fails, the error message remains readable?
that should be sufficient..

addmod....clog....PASS
compute_primes....discrete_log_bsgs....divrem2_precomp....PASS
PASS
PASS
...

from flint2.

You asked it to run all the test programs at the same time. So it did. I
can't imagine what else you were expecting it to do?

If you want the outputs to be sequential, run them sequentially!

Bill.

On 19 March 2014 00:28, jakobkroeker [email protected] wrote:

Why? It's not broken.

doh, probably you are right
the output gets scrambled (see below) but if a test fails, the error
message remains readable?
that should be sufficient..

addmod....clog....PASS
compute_primes....discrete_log_bsgs....divrem2_precomp....PASS
PASS
PASS
...

Reply to this email directly or view it on GitHubhttps://github.com//issues/54#issuecomment-38001793
.

from flint2.

If you want the outputs to be sequential, run them sequentially!

running tests sequentially is not mandatory to get the output atomic.

remark: the recent gnu make (v>4.0) provides an
--output-sync option, maybe that helps (have to update make before testing it)

from flint2.

If you want the outputs to be sequential, run them sequentially!

that takes far too long on a machine with several cores....

from flint2.