Re: [ecasound] Audio::Ecasound iam test failures with recent perl

Hi Joel, Kai and List,

(I'm the long lost Audio::Ecasound Perl module developer,
Joel recently discovered a bug which is detailed below)

I tested Audio::Ecaound with 2.4.4 and can replicate the corrupted
results from eci('int-cmd-list'). My test script is attached (test.pl)

The bug seems spookily dependent on previous use of other
libecasoundc functions. For example, the following works fine

$ perl ./test.pl | cat -A
last_type: S$
last_string_list_count: 188$

But if the BAD line is uncommented the output is:

$ perl ./test.pl | cat -A
last_type: S$
last_string_list_count: 188$
1: <ai-add^Z>$

Adding "unshift @_,1", for example, changes the corruption:

$ perl ./test.pl | cat -A
last_type: S$
last_string_list_count: 188$
1: <ai-add^Y>$

I replicated what Audio::Ecasound is doing in C and was able to
create a the same mess there (bug.c, attached)

$ gcc bug.c -o bug `libecasoundc-config --cflags --libs`
$ ./bug | perl -nle 'print if /[^\w\-\?]/' | cat -A
Bug test$
last_type: S$
count: 188$
?M-yM-|M-70M-@^F^H$
ai-addM-|^H$
ao-adda^H$
c-addv6^H$
c-liste^H$
cspM-7M-`M-GX^H$
cs-adde^H$
debug-M-^I^H$
esgM-7@email-addr-hidden-NM-^\^H$
fsrM-7pM-NM-^^^H$
getposs^H$
heM-|M-7M-hM-NM-#^H$
helpoM-OM-$^H$
qrM-|M-7M-^XM-PM-5^H$
quiteM-PM-6^H$
rewind-^H$
runiM-`M-PM-8^H$
rwnM-7M-xM-PM-9^H$
swM-|M-7^PM-QM-:^H$
setposs^H$
sttM-7XM-QM-=^H$
startsM->^H$
stopuM-yM-|M-7$

When eci_command("int-cmd-list") is run twice, the second run is
corrupted for some reason. At least that's the way it looks at the
moment, modulo my cruddy C and aged understanding on libecasoundc.

Is this a libecasoundc bug, or am I using it incorrectly?

Brad

-- 
  A man who will criticize you openly carries no connivance.
                      -- Hagakure http://bereft.net/hagakure/

_______________________________________________
Ecasound-list mailing list
Ecasound-list@email-addr-hidden
https://lists.sourceforge.net/lists/listinfo/ecasound-list