Hi Nikhil -
I'm sorry I didn't notice this earlier - but your program
works in this way because the channel.read functions taking
in variables return true if they read something and false
otherwise. That's because there are many cases in which
you don't know how long your file is, so you want something
like
while ffr.read(x) {
do something with x
}
The read functions taking in types already return the read
value, and so they can't return a separate 'bool' indicating
whether or not the read is successful. These variants halt
the program if the read was not successful.
While the language specification describes these read methods,
they don't describe the return types. I'll see if I can fix that.
In the mean time, the best way to learn about the I/O capabilities
is by looking over the example primers code in
examples/primiers/fileIO.chpl in this case.
I believe you will find that this behavior has nothing
to do with multiple locales (ie, if you replaced the
outer loop with forj in 1..2 it would work the same).
Cheers,
-michael
On 03/28/2014 09:41 PM, Nikhil Padmanabhan wrote:
> Hello,
>
> I'm a newbie to Chapel, so apologies if I missed seeing this issue
> somewhere...
>
> I just came across an edge case that surprised me while reading in a text
> file; see the attached code and input data. Very quickly, I was trying to get
> the various locales to read in consecutive chunks of a text file, and found
> that on locales other than the one I opened the file on, it lets me read past
> the file without an error.
>
> This does not happen if I replace the
> read(x,y,z);
> with
> (x,y,z) = read(real, real, real);
>
> I'm compiling the code with
> chpl -o test_reading test_reading.chpl
>
> CHPL_COMM=gasnet
> CHPL_COMM_SUBSTRATE=mpi
>
> This is on an OS X 10.8.5 system, with gcc 4.7. The command I ran, and the
> output is also attached.
>
> An unrelated comment :
>
> The writef statement triggers the warnings below while compiling --- I
> realize this is not yet in the language spec though.
>
> Thanks in advance!
> -- Nikhil
>
> In file included from /tmp/chpl-npadmana-30889.deleteme/_main.c:39:0:
> /tmp/chpl-npadmana-30889.deleteme/IO.c: In function '_format_reader':
> /tmp/chpl-npadmana-30889.deleteme/IO.c:6797:7: warning: passing argument 3 of
> 'qio_conv_parse' from incompatible pointer type [enabled by default]
> In file included from
> /Users/npadmana/myWork/chapel/chapel-1.8.0/runtime//include/qio/qio-all.h:12:0,
> from
> /Users/npadmana/myWork/chapel/chapel-1.8.0/runtime//include/chplio.h:16,
> from
> /Users/npadmana/myWork/chapel/chapel-1.8.0/runtime//include/stdchpl.h:33,
> from /tmp/chpl-npadmana-30889.deleteme/chpl__header.h:31,
> from /tmp/chpl-npadmana-30889.deleteme/_main.c:1:
> /Users/npadmana/myWork/chapel/chapel-1.8.0/runtime//include/qio/qio_formatted.h:804:8:
> note: expected 'size_t *' but argument is of type '_ref_uint64_t'
> In file included from /tmp/chpl-npadmana-30889.deleteme/_main.c:39:0:
> /tmp/chpl-npadmana-30889.deleteme/IO.c: In function '_extractMatch':
> /tmp/chpl-npadmana-30889.deleteme/IO.c:13103:5: warning: passing argument 6
> of 'qio_channel_read_string' from incompatible pointer type [enabled by
> default]
> In file included from
> /Users/npadmana/myWork/chapel/chapel-1.8.0/runtime//include/qio/qio-all.h:12:0,
> from
> /Users/npadmana/myWork/chapel/chapel-1.8.0/runtime//include/chplio.h:16,
> from
> /Users/npadmana/myWork/chapel/chapel-1.8.0/runtime//include/stdchpl.h:33,
> from /tmp/chpl-npadmana-30889.deleteme/chpl__header.h:31,
> from /tmp/chpl-npadmana-30889.deleteme/_main.c:1:
> /Users/npadmana/myWork/chapel/chapel-1.8.0/runtime//include/qio/qio_formatted.h:375:7:
> note: expected 'ssize_t * restrict' but argument is of type '_ref_int64_t'
>
>
> ---------------------------------
> Nikhil Padmanabhan
> [email protected] <mailto:[email protected]>
------------------------------------------------------------------------------
_______________________________________________
Chapel-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/chapel-users
