MIDI::ALSA 1.20 fails to install on Ubuntu 15.10 via CPAN

[Andy]

I just tried to install MIDI::ALSA on a freshly installed Ubuntu 15.10
PC (yes I know Ubuntu support for 15.10 stopped recently but that's not
relevant for this issue), but it fails during the testing phase on Test 35:

ok 29 - skipping note_on event output
ok 30 - skipping note_off event output
ok 31 - skipping disconnectfrom()
ok 32 - skipping SND_SEQ_EVENT_PORT_UNSUBSCRIBED event
ok 33 - skipping disconnectto()
ok 34 - connectto(2,'testpl pid=4697:1') connected to myself by name
not ok 35 - received an event from myself
# Failed test 'received an event from myself'
# at test.pl line 301.
# alsaevent=$VAR1 = [11,1,0,1,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];
# correct =$VAR1 = [11,1,0,0,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];
ok 36 - latency was 9 microsec
ok 37 - disconnectfrom(1,128,2)
ok 38 - status() reports running
ok 39 - status() reports time = 2.805671 not 2.80858683586121
ok 40 - status() reports time = 3.808081 not 3.81115889549255
ok 41 - stop() returns success
[...]
# Looks like you failed 1 test of 57.
Makefile:1010: recipe for target 'test_dynamic' failed
make: *** [test_dynamic] Error 1
PJB/MIDI-ALSA-1.20.tar.gz
/usr/bin/make test -- NOT OK

What could be the cause of this error?

Many thanks in advance for any help.

[Andy]

Post by Andy
I just tried to install MIDI::ALSA on a freshly installed Ubuntu 15.10
PC (yes I know Ubuntu support for 15.10 stopped recently but that's not
ok 29 - skipping note_on event output
ok 30 - skipping note_off event output
ok 31 - skipping disconnectfrom()
ok 32 - skipping SND_SEQ_EVENT_PORT_UNSUBSCRIBED event
ok 33 - skipping disconnectto()
ok 34 - connectto(2,'testpl pid=4697:1') connected to myself by name
not ok 35 - received an event from myself
# Failed test 'received an event from myself'
# at test.pl line 301.
# alsaevent=$VAR1 = [11,1,0,1,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];
# correct =$VAR1 = [11,1,0,0,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];
ok 36 - latency was 9 microsec
ok 37 - disconnectfrom(1,128,2)
ok 38 - status() reports running
ok 39 - status() reports time = 2.805671 not 2.80858683586121
ok 40 - status() reports time = 3.808081 not 3.81115889549255
ok 41 - stop() returns success
[...]
# Looks like you failed 1 test of 57.
Makefile:1010: recipe for target 'test_dynamic' failed
make: *** [test_dynamic] Error 1
PJB/MIDI-ALSA-1.20.tar.gz
/usr/bin/make test -- NOT OK
What could be the cause of this error?
Many thanks in advance for any help.

I just noticed the same test 35 fails on my old Mandriva box too, so it
seems a general problem, not related to a specific distro.

[Peter Billam]

Post by Andy

Post by Andy
I just tried to install MIDI::ALSA on a freshly installed Ubuntu
ok 34 - connectto(2,'testpl pid=4697:1') connected to myself by name
not ok 35 - received an event from myself
# Failed test 'received an event from myself'
# at test.pl line 301.
# alsaevent=$VAR1 = [11,1,0,1,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];
# correct =$VAR1 = [11,1,0,0,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];
ok 36 - latency was 9 microsec
[...]
# Looks like you failed 1 test of 57.
What could be the cause of this error?

I just noticed the same test 35 fails on my old Mandriva box too,
so it seems a general problem, not related to a specific distro.

OK, that's my module... I just got back from hospital,
and am running on half-power at the moment :-( Anyway:

I can assure you that the reported error ( 11,1,0,1 <==> 11,1,0,0 )
is meaningless. It's a bug in test.pl.
Which MIDI-ALSA version were you using ?
AFAICS in MIDI-ALSA-1.20 the bug is fixed, and that should test OK...

Regards, Peter

[Andy]

Post by Peter Billam

Post by Andy

Post by Andy
I just tried to install MIDI::ALSA on a freshly installed Ubuntu
ok 34 - connectto(2,'testpl pid=4697:1') connected to myself by name
not ok 35 - received an event from myself
# Failed test 'received an event from myself'
# at test.pl line 301.
# alsaevent=$VAR1 = [11,1,0,1,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];
# correct =$VAR1 = [11,1,0,0,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];
ok 36 - latency was 9 microsec
[...]
# Looks like you failed 1 test of 57.
What could be the cause of this error?

I just noticed the same test 35 fails on my old Mandriva box too,
so it seems a general problem, not related to a specific distro.

OK, that's my module... I just got back from hospital,

I'm sorry to hear that, I sincerely hope that you recover completely
soon (not just because of MIDI-ALSA ;) ).

Post by Peter Billam
I can assure you that the reported error ( 11,1,0,1 <==> 11,1,0,0 )
is meaningless. It's a bug in test.pl.

Yep, that's what I thought, but it looks bad when trying to get other
people to install MIDI-ALSA.

Post by Peter Billam
Which MIDI-ALSA version were you using ?
AFAICS in MIDI-ALSA-1.20 the bug is fixed, and that should test OK...

It's in 1.20 downloaded from CPAN, maybe you fixed it in your code at
home, but never got round to update it on CPAN?

By the way, I got MIDI-ALSA into Debian, although they are deciding how
to handle the kernel module requirements of 'make test' (virtmidi), so
it hasn't become a package yet. See here:
https://anonscm.debian.org/git/pkg-perl/packages/libmidi-alsa-perl.git

If you would like to suggest a way forward for them then please tell
them on the debian-perl mailing list: debian-***@lists.debian.org
See the thread here:
https://lists.debian.org/debian-perl/2016/10/msg00033.html

Best regards,
Andy

[Peter Billam]

Post by Andy

Post by Peter Billam
I can assure you that the reported error ( 11,1,0,1 <==> 11,1,0,0 )
is meaningless. It's a bug in test.pl.

Yep, that's what I thought, but it looks bad when trying to get
other people to install MIDI-ALSA.

The problem is that sometimes alsa assigns queue ids starting
from 0 and sometimes from 1 ; I don't know the background. On my
debian jessie, test.pl runs fine, which makes it hard to test.

Post by Andy

Post by Peter Billam
Which MIDI-ALSA version were you using ?
AFAICS in MIDI-ALSA-1.20 the bug is fixed, and that should test OK...

It's in 1.20 downloaded from CPAN, maybe you fixed it in your code
at home, but never got round to update it on CPAN?

Try MIDI-ALSA-1.21 then ...

Post by Andy
By the way, I got MIDI-ALSA into Debian, although they are deciding
how to handle the kernel module requirements of 'make test'
https://anonscm.debian.org/git/pkg-perl/packages/libmidi-alsa-perl.git

Thanks for your work !

Regards, Peter

[Andy]

Post by Peter Billam
The problem is that sometimes alsa assigns queue ids starting
from 0 and sometimes from 1 ; I don't know the background. On my
debian jessie, test.pl runs fine, which makes it hard to test.

Wouldn't it be possible to ignore the queue id in the test and only
compare all other parameters?

Post by Peter Billam

Post by Andy

Post by Peter Billam
Which MIDI-ALSA version were you using ?
AFAICS in MIDI-ALSA-1.20 the bug is fixed, and that should test OK...

It's in 1.20 downloaded from CPAN, maybe you fixed it in your code
at home, but never got round to update it on CPAN?

Try MIDI-ALSA-1.21 then ...

Tried it, but still get the error as before:

not ok 35 - received an event from myself
# Failed test 'received an event from myself'
# at test.pl line 301.
# alsaevent=$VAR1 = [11,1,0,1,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];
# correct =$VAR1 = [11,1,0,0,'2.8',[128,2],[128,1],[0,0,0,0,0,99]];

Post by Peter Billam

Post by Andy
By the way, I got MIDI-ALSA into Debian, although they are deciding
how to handle the kernel module requirements of 'make test'
https://anonscm.debian.org/git/pkg-perl/packages/libmidi-alsa-perl.git

Thanks for your work !

You're welcome, but I'm also doing it to make my life easier as it's
much easier to install MIDI::ALSA from a repo than having to build it
every time.

The above error in 'make test' also slightly further complicates things
as it makes a 'cpan -fi MIDI::ALSA' necessary rather than just a
straightforward 'cpan MIDI::ALSA'.

By the way, thanks for >>your<< work. Without MIDI::ALSA I wouldn't have
been able to write any of the MIDI programs that I wrote in the past 4
years.

Best regards,
Andy

[Peter Billam]

Post by Andy
Wouldn't it be possible to ignore the queue id in the test and
only compare all other parameters?

Indeed it was, but I forgot to re-make the tarball :-(
This would not happen if I was well and running on full energy.
Apologies.

So try MIDI-ALSA-1.22 then (when it hits the mirrors) ...

Regards, Peter