Re: [GLE-devel] Embedding GLE?

[<po...@di...>]

Hello Jan,

First off, thanks for the comprehensive reply!

I'm working on an application that frequently has to display fairly complex diagrams (schematic-like overviews with status indicators, graphs and the like) representing the current status or result of a process. The "boilerplate" for those diagrams has to be user-controllable, i.e. is defined by a superuser as part of the configuration for a particular site; the actual data (positions, numbers, data etc) is then filled in at run time.

The GLE scripting language seemed quite ideally suited for that- and actually I already tried with some success what you described in (1), creating a GLE script from a template, running the GLE process on it, loading the generated image (PNG, say) and displaying it. 

That works, but it's unwieldy. There is the performance penalty, which verges on being prohibitive; also creating subprocesses is undesirable for various reasons. 

So what you suggested in (2) sounds good and that small sample you mentioned would indeed be very helpful. I won't completely get around temp files since I'll have to plot graphs now and then, but that shouldn't be a problem for now. 

thank you again,

Best regards,
Dietmar


----- Original Message -----
From: jan...@st...
To: po...@di...
Date: 19.02.2013 20:20:39
Subject: Re: [GLE-devel] Embedding GLE?


> 
> Dear Dietmar,
> 
> It depends on what you want to accomplish. Can you eleborate a 
> bit on that?
> 
> If you want to create a GLE script dynamically and then renter it 
> to screen, then there are basically two possibilities.
> 
> (1) Write the script to a temporary file and run the GLE 
> executable on this to produce a PNG. Next, show this PNG on 
> screen. For this, you will need some framework that can display 
> PNGs, such as QT. If you use the command line option -cairo, then 
> you don't need GhostScript for this. The advantage is that you 
> have a very simple interface.
> 
> (2) Link to the GLE library and do the above with function calls 
> to the library. The advantage of this is that the temporary files 
> are avoided to some extend. But, if your script contains LaTeX 
> expressions (real LaTeX, not GLEs built in simple TeX), then GLE 
> will create temporary files anyway because latex can only work on 
> files. If your drawings use graphs with data sets, then you will 
> also need (temporary) files for the data sets. The advantage of 
> this approach is that it may be faster and that you don't have to 
> worry about creating temporary files and corresponding error 
> handling.
> 
> In (2) you will also need to be able to show PNGs. GLE will write 
> the PNG to a memory buffer (not a file in this case) and then you 
> need some other library to show this PNG memory buffer on screen. 
> If you use Cairo in your application, then it would be possible 
> to get around this, but in QGLE we have used PNG as the 
> communication protocol between GLE and QT as it is easier than to 
> define our own (uncompressed) format.
> 
> One problem with (2) is that I'm afraid that no one has tried it 
> so far. It has come up for discussion a couple of times, but I 
> don't know of any cases were it was done completely.
> 
> If you are interested in (2), then please let me know and I will 
> write a small example program.
> 
> Best Regards,
> 
> Jan