Thank you David, that is exactly what I see. I invoked the “command” always without “+”. However, this is only a temporary effect during addition of command specification +. I think the definition of “command” and “+command” should not differ in normal circumstances.
I’m not sure what you mean here. Maybe a simple example to discuss about it?
Actually, I had a “command” in update299.gmic which in fact did not need the selection of any image selection list. With definition of a new “+command” e.g. in my .gmic I try to change that, possibly with further changes.
Finally I would replace the “command” definition by “+command” definition, not adding the latter. It might be difficult to have both definitions in mind.
After a few days of experimenting with that, what I can say is :
- If you have a command that basically inserts a new image in the list, without really needing a selection, then it’s probably good to have only a
+commandspecialization defined. It’s the case for instance, with the commandclut, or commands that insert a new 3D object (e.g.torus3d, pyramid3d), and the different commands in theinpux_XXXfamily. - Otherwise, sticking with regular
commanddefinitions is OK.
It’s actually rare that both command and +command need to be defined. It’s the case only when a command requires an image selection (with possibly a lot of images) but does not modify it (e.g. command psnr). In that case, the +command specialization is useful to avoid useless image copies in memory.
- 2021/09/02 : Release of 2.9.9 (last one before 3.0.0!).
1 Like
What’s on plan for 3.0.0 and above? Or should you not tell the spoilers?
What I really want as a new feature for 3.0.0 is at least one filter based on the machine learning library I’ve started writing for G’MIC. Apart from that, I’d like to polish the code to get more optimizations and eradicate bugs (if possible 
).
It doesn’t look very ambitious at a first glance, but the machine learning stuff will be really one big thing to achieve.
1 Like
ML alone is big. Thanks for introducing it. I and others have winked too many times. 
1 Like
Maybe then you can simulate me, David. Unfortunately, it would likely go crazy in the process. lololololololol
Is this a bug, or am I doing something wrong?
gmic v -99 in.png fx_symmetric_shape2d 5,42.0,17.42,\"42.0,17.42\",12.0,13.86,\"12.0,13.86\",10.0,37.13,1.0,2.70,23.0,26.25,12.0,14.15,42.0,-6.45,2.0,6.73,1,152,214,81,100 o out_000.png
[gmic] *** Error in ./fx_symmetric_shape2d/*repeat/*local/ *** Command ‘if’: Invalid argument '!0 && ': Unrecognized item ‘&’ in expression ‘!0 &&’.
EDIT: I’m using Windows cli version 2.9.9 (pre-release #210815) (and I updated it too…)
That is a kind of bug yes.
I’ve fixed it, and updated the filter definition file.
$ gmic up
should let you get the fix.
That is a kind of bug yes.
I’ve fixed it, and updated the filter definition file.
$ gmic up
should let you get the fix.
Thank you! It works perfectly now 
1 Like
I’m sure I asked this a long time ago, but what’s a good way to search (in the CLI) among the defined commands for a particular word or phrase? For example, find commands containing the word “blur”. If there was a way to list all currently defined commands (is there?) we could pass it through grep etc.
2 Likes
gmic parse_cli print,blur
gives
$ gmic parse_cli print,blur
[gmic]-0./ Start G'MIC interpreter.
[gmic]-0./ Parse '#@cli' command(s) 'blur' and output in 'print' mode.
blur
blur_angular
blur_bloom
blur_linear
blur_radial
blur_selective
blur_x
blur_xy
blur_xyz
blur_y
blur_z
deblur
deblur_goldmeinel
deblur_richardsonlucy
deblur_upscale_goldmeinel
deblur_upscale_richardsonlucy
frame_blur
gcd_blur_masked
gcd_blur_texture
mapblur
rep_blur_splinter
rep_fragment_blur
rep_frblur
rep_splinter_blur_convolve_map
[gmic]-0./ End G'MIC interpreter.
(list of commands output on stdout)
2 Likes
Really helpful. I think that could become a separate command!
cmd_find : parse_cli print,$1
or
cmd_find :
l[] it $_path_rc/update${_version}.gmic it $_path_user a y parse_cli print,$1 rm endl
Another thing, is there a way to activate help printed for custom commands beginning with “_”?
Do you have problem with procedure entry point located in dynamic link library Qt5Gui.dll and 2.10.28?
Yes, just tested today and the plug-in doesn’t work anymore with latest GIMP 2.10.28.
No idea why this happens.
If someone can explain a solution…
@David_Tschumperle
(réponse en français)
Bonjour,
L’exécutable ‘gmic_gimp_qt.exe’ des archives de gmic.eu fonctionne correctement lorsqu’il est mis dans un autre environnement de DLL (qt,etc.).
Je pense que pour éviter un conflit de DLL et une amélioration des distributions ces options sont envisageables:
- Fournir l’ensemble des DLL nécessaires au fonctionnement de ‘gmic_gimp_qt.exe’ sauf les DLL propres à GIMP. Par exemple, mes archives regroupent les DLL nécessaires au fonctionnement de ‘gmic.exe’ et ‘gmic_qt.exe’ et je regroupe les 3 exécutables pour faire une seule distribution.
- Mettre à jour l’environnement de compilation.
Merci :o)
1 Like
-
I’m trying to reduce the number of .dll embedded in the G’MIC plug-in installer or the .zip archive.
.dllfiles are intended to be shared between different software. I know we are on Windows and theory and practices are two different things on this platform, but at least I’d want to respect this rule as much as possible (and I don’t want installers that takes 300Mo of .dll files). -
I’ve updated the compiling environment with
pacman -Syuand it went all the way to end, without issues. But when I’ve recompiled the plug-in binary, I got the same error message.
1 Like