[Scilab-users] Cannot understand scilex behavior when piping

[Antoine Monmayrant]

Hi,

I'm not sure I'll be of any help, as I'm using Scilab on a linux machine.
Is there an equivalent of scilex.exe on linux?
(Is it scilab-adv-cli or scilab-cli?)
If yes, I can try to run a test case under Linux using the corresponding 
equivalent to see if it's a problem with pipes or pipes on Windows.

Antoine

Le 10/01/2015 03:51 PM, Laurent Bonaventure a écrit :
> Hello.
>
> I'm still trying to understand the difference in behavior when opening 
> a graphic windows, between:
> - opening it directly from Scilex
> - opening it from Scilex, but when Scilex stdin is piped
>
> I managed to narrow down the behavior to the opening of a new graphic 
> window by java.
>
> In Scilex (Windows 7+Scilab 5.5.2), if you type:
>
> jimport org.scilab.modules.graphic_objects.builder.Builder
> jimport org.scilab.modules.graphic_objects.CallGraphicController
> p=jinvoke(Builder,"createNewFigureWithAxes")
> jinvoke(CallGraphicController,"setGraphicObjectProperty",p,357,%f) 
> //__GO_MENUBAR_VISIBLE__, false
> jinvoke(CallGraphicController,"setGraphicObjectProperty",p,355,%f) 
> //__GO_TOOLBAR_VISIBLE__, false
> jinvoke(CallGraphicController,"setGraphicObjectProperty",p,358,%f) 
> //__GO_INFOBAR_VISIBLE__, false
> jinvoke(CallGraphicController,"setGraphicObjectProperty",p,357,%t) 
> //__GO_MENUBAR_VISIBLE__, true
> jinvoke(CallGraphicController,"setGraphicObjectProperty",p,355,%t) 
> //__GO_TOOLBAR_VISIBLE__, true
> jinvoke(CallGraphicController,"setGraphicObjectProperty",p,358,%t) 
> //__GO_INFOBAR_VISIBLE__, true
>
> then you get an empty graphic window (java), and when you select 
> Edit/Axes Properties, the Tk "graphic editor" window shows up 
> immediately.
>
> If you send the same commands to Scilex "through a pipe" (eg: cat | 
> scilex) (assuming you have "cat" on your windows), then the graphic 
> window appears normally, but the Tk window don't show up when you 
> select Edit/Axes Properties.
>
> Strangely, if you then close the graphic window and then reopen it (by 
> sending the above commands again), the Tk windows shows up (too late).
>
> Could someone give me a hint about what's going on here? Why isn't Tk 
> interacting correctly with the java graphic window in the "pipe" case, 
> while everything goes normally in the "non-pipe" case?
>
> Thanks
>
> Laurent Bonaventure
>
> Le 20/09/2015 04:41, Laurent Bonaventure a écrit :
>> Hello.
>>
>> I'm still trying to pipe instructions into scilex (Windows 7, Scilab 
>> 5.4).
>>
>> When I send a plot command, scilex opens the graphics window and draw my
>> polyline -> Great !
>>
>> Then I try to open the graphics editor from within the graphics window
>> (e.g. Edit/Axes properties) : nothing happens.
>>
>> Then I close the graphics window.
>>
>> Then I send the plot command again. And poof! The graphics windows opens
>> AND the long awaited graphics editor too. Except that nothing works 
>> in it.
>>
>> I guess it's some sort of handlers problem: the graphics window cannot
>> communicate correctly with the scilab kernel and get the correct
>> handlers to the figure.
>>
>> What's more strange, is that the behaviour is correct when I start
>> scilex from within an interactive console (cmd)...
>>
>> Steps to reproduce (you need some utility like cat from gnuwin32, or any
>> way to pipe commands to scilex) :
>>
>> (Incorrect behaviour, with a pipe)
>> <from an interactive cmd.exe>
>> cat | scilex
>> x = 0:0.1:10
>> plot(x,cos(x))
>> <the graphic window opens>
>> <click on Edit/Axes properties>
>> <nothing happens>
>> <close the graphic window>
>> plot(x,cos(x))
>> <the graphic window opens, and the graphic editor too (but it isn't
>> functional)>
>>
>> (Correct behaviour, without pipe)
>> <from an interactive cmd.exe>
>> scilex
>> x = 0:0.1:10
>> plot(x,cos(x))
>> <the graphic window opens>
>> <click on Edit/Axes properties>
>> <the graphic editor shows up immediately and is functional>
>>
>> I don't understand how scilex can see any difference between both
>> approaches, and why it would have a different behaviour based on that
>> difference. From its point of view, there shouldn't be any difference.
>>
>> Ah! And the -nw switch doesn't change anything... You get exactly the
>> same behaviour in both cases.
>>
>> Can someone give me some sort of solution, or pointers, or even give me
>> an idea for a workaround. I need to pipe things in scilex, and I need
>> the first approach to work correctly...
>>
>> Thank you.
>> Laurent
>
>
> _______________________________________________
> users mailing list
> users at lists.scilab.org
> http://lists.scilab.org/mailman/listinfo/users
>