Thanks, I'll keep an eye on that PR.
One other thing that is also missing is the possibility of returning the
command status like Matlab does. In my workaround above I force it to be
ignored (otherwise it may abort the function execution) but we than have no
info of its value. Right, one can do a
success(cm)
but this implies running the command twice ... and `cmd` may do a heavy job
sábado, 12 de Setembro de 2015 às 15:51:40 UTC+1, Miguel Bazdresch escreveu:
>
> See this PR: https://github.com/JuliaLang/julia/pull/12807/files
>
> This is a change to `open` that would allow capturing STDERR the way you
> want. This change won't make it into v0.4, but the code in there could
> point you towards a solution. I hope to have some time soon to dive into
> this (I need to implement the same thing in Gaston).
>
> -- mb
>
> On Sat, Sep 12, 2015 at 10:47 AM, J Luis <[email protected] <javascript:>>
> wrote:
>
>> Right, that's indeed what is (very sadly) happening. It turned out that
>> the stderr output is the part that I really need in this case.
>> The only working solution I found was to do
>>
>> run(pipeline(ignorestatus(cmd), stdout=DevNull, stderr="errs.txt"))
>> t = readall("errs.txt");
>>
>> but it's annoying having to write a disk file and read its contents.
>>
>> Had some hope on this one, but no way too
>>
>> julia> readall(run(pipeline(ignorestatus(cm), STDOUT, STDOUT)))
>> Image Difference (RootMeanSquaredError):
>> Normalized Absolute
>> ============ ==========
>> Red: 0.3709618690 24311.0
>> Green: 0.3611391197 23667.3
>> Blue: 0.2638682470 17292.6
>> Total: 0.3354900017 21986.3
>> C:/programs/GraphicsMagick/gm.exe compare: image difference exceeds limit
>> (0.33549 > 0.001).
>> ERROR: MethodError: `readall` has no method matching readall(::Void)
>>
>> and this one completely hangs the REPL giviing me no choice but killing
>> the shell window
>>
>> rd,wr=redirect_stderr();
>> run(pipeline(ignorestatus(cmd), stdout=DevNull, stderr=rd))
>>
>> now if try to read from 'rd'
>>
>> readall(rd) # ---> IceAge
>>
>>
>> sábado, 12 de Setembro de 2015 às 15:35:39 UTC+1, Miguel Bazdresch
>> escreveu:
>>>
>>> What could be happening is that `gm.exe` is printing that message to
>>> STDERR, which is not captured by `readall()`. In theory, the new pipeline
>>> infrastructure should let you capture STDERR, but I haven't had the time to
>>> figure out exactly how. Maybe somebody who knows will chime in.
>>>
>>> -- mb
>>>
>>> On Fri, Sep 11, 2015 at 9:17 PM, J Luis <[email protected]> wrote:
>>>
>>>> Found a ignorestatus function. Better, but not yet good enough. It
>>>> still prints a message even if using a suppression ';'
>>>>
>>>> julia> readall(ignorestatus(cm));
>>>> C:/programs/GraphicsMagick/gm.exe compare: image difference exceeds
>>>> limit (0.336711 > 0.001).
>>>>
>>>>
>>>> sábado, 12 de Setembro de 2015 às 02:00:25 UTC+1, J Luis escreveu:
>>>>>
>>>>> Thanks. It turned out my main error was that I was build the Cmd
>>>>> object by first creating the command as a string and after wrapping it in
>>>>> back ticks.
>>>>>
>>>>> But I continue in troubles. Right, I can run the command, a
>>>>> GraphicsMagic image comparison command, but when that comparison say the
>>>>> two images are different gm.exe returns an error code != 0 and Julia
>>>>> interprets it a command error. But it isn't, it only means the two images
>>>>> are different. As a consequence the julia function where this happens
>>>>> aborts (an example in REPL bellow).
>>>>>
>>>>> I tried with a try catch but not even that prevented the function
>>>>> abortion. How can I get out of this one?
>>>>>
>>>>> julia> readall(cm)
>>>>> C:/programs/GraphicsMagick/gm.exe compare: image difference exceeds
>>>>> limit (0.33549 > 0.001).
>>>>> ERROR: failed process: Process(`C:/programs/GraphicsMagick/gm.exe
>>>>> compare -density 200 -maximum-error 0.001 -highlight-color magenta
>>>>> -highlight-style assign -metric rmse -file V:/example_02.png
>>>>> C:/progs_cygw/GMTdev/gmt5/branches/5.2.0/doc/examples/ex02/
>>>>> example_02.ps V:/example_02.ps`, ProcessExited(1)) [1]
>>>>> in pipeline_error at process.jl:548
>>>>>
>>>>> sábado, 12 de Setembro de 2015 às 01:17:57 UTC+1, Simon Kornblith
>>>>> escreveu:
>>>>>>
>>>>>> readall(`cat test`) or similar
>>>>>>
>>>>>> On Friday, September 11, 2015 at 7:56:43 PM UTC-4, J Luis wrote:
>>>>>>>
>>>>>>> Ok, I've spend about an hour around "run" "open", "run(pipeline(..."
>>>>>>> but no way.
>>>>>>> In Matlab I would do
>>>>>>>
>>>>>>> [status, cmdout] = system(cmd);
>>>>>>>
>>>>>>> but in Julia the most a can reach is to run the command
>>>>>>>
>>>>>>> com = "C:/programs/GraphicsMagick/gm.exe compare -density 200 ...
>>>>>>>
>>>>>>> run(`com')
>>>>>>>
>>>>>>> but I need the result of that execution.
>>>>>>> How to?
>>>>>>>
>>>>>>> Thanks
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>
>
