Re: НА: НА: НА: Smoke [ebcdic] v5.21 .8-419-gaa57d5a FAIL(X) os/390 23.00 (2817/)

On 02/12/2015 11:05 PM, Yaroslav Kuzmin wrote:
>> If I had been thinking, I would have realized that.  Attached is a
>> revised file, that I hope works for you.  It won't compile on my system
>> since things are different here.
>
>> The difference is that your system has an element in the DIR structure
>> called dd_fd which should contain the same information as what dirfd()
>> returns.
>
>> If it doesn't compile, maybe you could tweak it so it does.
>
>
> (00:38) RS12 : PDKUZM | ~/ussport/perl/test :>
> (00:38) RS12 : PDKUZM | ~/ussport/perl/test :> c99 -qlanglvl=extended:extc89:extc99 -qlongname -qxplink -qdll  -DOEMVS  -D_XOPEN_SOURCE_EXTENDED -D_ALL_SOURCE -D_POSIX_SOURCE=1 -c stat.c -o stat.o
> (00:38) RS12 : PDKUZM | ~/ussport/perl/test :> c99 -qxplink -qdll -o stat stat.o
> (00:38) RS12 : PDKUZM | ~/ussport/perl/test :> ./stat
> fstat: EDC5113I Bad file descriptor. (errno2=0x053A0090)
> (00:38) RS12 : PDKUZM | ~/ussport/perl/test :>
>
> I do not know why he gives out "Bad file descriptor" :)

Is there IBM support that you could report this to?  I searched the 
internet and didn't find anything.
>
> But to fill the structure stat for directory can use function stat :
>
> output attached file
>
> (00:56) RS12 : PDKUZM | ~/ussport/perl/test :>
> (00:59) RS12 : PDKUZM | ~/ussport/perl/test :> c99 -qlanglvl=extended:extc89:extc99 -qlongname -qxplink -qdll  -DOEMVS  -D_XOPEN_SOURCE_EXTENDED -D_ALL_SOURCE -D_POSIX_SOURCE=1 -c stat.c -o stat.o
> (00:59) RS12 : PDKUZM | ~/ussport/perl/test :> c99 -qxplink -qdll -o stat stat.o
> (00:59) RS12 : PDKUZM | ~/ussport/perl/test :> ./stat
>   :> 3
> fstat: EDC5113I Bad file descriptor. (errno2=0x053A0090)
> Success : stat

We know we could use stat, but Perl tests to see if fstat should work; 
it saves cycles in some situations, if it is available.  I'll override 
it for now to say it isn't available.

I have now pushed a new version for you to smoke which should fix the 
new bom failure, and should skip the failing stat failure.  It also has 
changes for the dumper.t failure, though neither of these .t's will pass 
as-is.  If dumper.t fails before it gets to test number 379, please run 
it individually by hand and send me the output, as you did before.  379 
and some later tests are expected to fail, and I know what to do to get 
them to pass; I just haven't had a chance to fix them.

By the way, your area was in the news, for getting blue snow.