Please read:
http://www.blender3d.org/cms/Imaging.834.0.html
Or in short:
- adding MultiLayer Image support
- recoded entire Image API
- better integration of movie/sequence Images
Was a whole load of work... went down for a week to do this. So, will need
a lot of testing! Will be in irc all evening.
Bug #4484:
http://projects.blender.org/tracker/?func=detail&aid=4484&group_id=9&atid=125
A recent update to Draw.c made the button attr setter function not exit
where it should for one of the cases, when new and old strings had
same length (code was missing a "return 0;" line in one of the if
cases). Should work fine now, I don't get the error under linux anymore.
Also saw in the same function that a py object could be created but
was not being decref'ed in two other 'if' cases.
Thanks elbarto for reporting the bug.
Window.py added note about the view offset being flipped.
editview.c Fly mode timed redraws so motion is the same speed erespective of scene complexity.
documented a workaround for python drawinf pixel aligned text in teh 2d view
Added links from Draw.Test to the OpenGL functions that set the location and color.
implicit declarations, redundant redeclarations, missing initializers,
nested externs and other cruft.
Cleaned up includes and moved extern _Type decls from Types.h into
Types.c since that is the only place where they are needed now.
Did not touch Ipo.[ch] since work is on-going there.
PupBlock method. This wraps the "clevernumbut" code to allow scripters to use popup blocks for user input instead of a sequence of multiple different popups.
See the blend file for a comprehensive test and example file.
PyDict_SetItemString() with objects that were not properly decrefed
afterwards. Due to the number of places this was used, I added a
wrapper EXPP_dict_set_item_str() to gen_utils.c to handle it.
This started as a scriptlink bug, due to how many times scripts were
being executed I think it just magnified how bad the memory leak in
BPy was. Animating the blend attached with this bug report would cause
memory to grow by about 3MB for every 280 frames. After the patch,
memory did not appear to grow at all (or at least not noticably using
Unix's ps and top utils).
Since many of the PyDict_SetItemString() calls were in initialization
modules I think my tests executed most of the changed code, but would
appreciate script users really giving feedback.
- Mostly this cleans up the #includes and header files in the python project.
- Warning fixes are mostly casting issues and misc fixes. General warning clean up.
- #include Python.h MUST come as the first include to avoid the POSIX redefine warning in the unix makefiles
- fno-strict-aliasing flag added to makefile to fix a unavoidable type punning warning in types.c
This fixes the active object in place when orbiting the view.
Choppy 15fps demo can be seen there: http://www.elysiun.com/~theeth/bf/around_active.html
Image Memory Grabage Collection
This adds memory handling to the image code. An image is tagged each time it is used.
During a collection cycle (frequency of cycles is user defined), if an image is older
than a user defined limit, its buffer gets deallocated. This also applies to gl memory buffers.
Images that are loading in GL memory needs to go through two time outs before being fully deallocated: the first time out deallocated the gl memorry, the second the buffer in ram.
Notes:
Image buffer loaded from python gets tagged as permanent upon load. That tag is removed when python stops using the image.
I might have missed some tagging spots, especially in the rendering pipeline. Someone with more knowledge about this code should check to be careful.
Tagging is done on every access, for rendering, this will probably be a performance hit. A scheme should be developped to only tag when the rendering is completed.
Collecting is called in draw_object, most likely not the best place to do it.
Safe from undo, since using undo deallocates memory anyway (like when loading a blend file with one currently opened)
Userpref DNA changes:
I've changed a couple of flagging variables from short to int. Some because they needed more space, others to keep SDNA happy.
Info window changes:
I've grouped a couple of buttons in aligned blocks and changed the color of mutually exclusive options to make them clearer.
Matt didn't do any changes on that in tuhopuu, so hopefully I'm not stepping on anyone's feet with this.
Also changed double constants into floats with f in a couple of places (mostly space.c) to make compiler happier.
- #2646 reported by Campbell: Python/Fileselector (moving from fileselector called by script to another space caused script to hang around open but not accessible)
http://projects.blender.org/tracker/?func=detail&atid=125&aid=2646&group_id=9
- #2676 reported by Wim Van Hoydonck: 2.37 python scripts gui: event 8 ignored (thanks Ton for discussing / pointing what to do, Ken Hughes for also working on a fix)
http://projects.blender.org/tracker/?func=detail&atid=125&aid=2676&group_id=9
- gui-less scripts with calls to progress bar inside fileselector callbacks didn't return to the previous space on exit (staying on Scripts win), requiring an event to do so (mouse movement, for example). Quick fix for now, will rework a little after 2.37a for a better alternative, not needing to move to the Scripts win at all.
- added syntax colors access to Window.Theme module.
Scripts:
- updates by Jean-Michel Soler: svg2obj (svg paths import), tex2uvbaker, fixfromarmature;
- updates by Campbell Barton: obj import / export, console;
- tiny: converted vrml97 export to unix line endings;
- updates in ac3d exporter, help browser, save theme.
Thanks all mentioned above.
- Made Blender.event var (previously only used by script links) hold ascii value -- where it applies -- of current event during events callback registered with Draw.Register(gui, events, button_events). Useful for gui scripts like Campbell's Python console. No problem using this var to hold the value, since in gui scripts it was not used (always None).
- Updated Window and Window.Theme with new theme vars and the Time space.
- Script links:
-- Added "Render" event for script links (runs twice, second time as "PostEvent", for clean-up actions). Now FrameChanged links don't run when a single pic is rendered.
-- Added "Enable Script Links" button in the script buttons tab. Now this bit gets saved in .blends along with the rest of G.f, so users can define per .blend if they are on or off by default. "blender -y" also disables all slinks as happened before with OnLoad ones only.
-- Other small changes in the script buttons tab:
When a link is added (button "new"), it becomes the active one for the window, no need to press a button to reach it.
Also, a pupmenu showing all available texts is shown when "new" is pressed, so users can choose a text w/o having to type. Cancel the popup to leave the string button empty (link exists, but has no script assigned). A pulldown would be better UI-wise, but it's kinda weird to show both scripts and normal texts (Blender doesn't differentiate them) in a script links pulldown. With a popup we can show only texts ending in ".py" (not done in this commit, need opinions) and if the script has no or another extension, case of many in old and current .blend's, there's still the string box for writing its name.
-- Implemented Ton's space handler script links:
Right now only for the 3d View, but it's trivial to add for others. There are two types: EVENT, to receive 3d View events from a chosen window and DRAW, to draw on the window. Ton's idea was to give scripts a controlled way to integrate better within Blender.
Here's how it works:
- scripts must have a proper header, like:
# SPACEHANDLER.VIEW3D.EVENT
and then they are shown in 3d View's View menu, "Space Handler Scripts" submenu. Check (mark, click on it) a script to make it active.
EVENT handlers should consult the Blender.event var to get the current event, which can be compared with values from the Draw module:
import Blender
from Blender import Draw
evt = Blender.event
if evt == Draw.AKEY:
print "a"
elif evt == Draw.LEFTMOUSE:
print "left mouse button"
else:
return # ignore, pass event back to Blender
Blender.event = None # tell Blender not to process itself the event
DRAW handlers are free to draw to their owner 3D View. OpenGL attributes and modelview and projection matrices are pushed before running the handler and poped when it finishes.
To communicate between EVENT and DRAW handler scripts we have the Blender.Registry module, as always.
Still need to code some nice example, which should also serve to test properly space handlers. Simple tests went fine.
- doc updates about the additions.
=======
Note: the UI part of the space handlers and script links is of course open for changes, I just tried to make it understandable. Probably we won't use the scriptlinks icon for "None Available" (check 3d View -> View -> Space Handler Scripts), though it hints at what space handlers are. The tooltips may not be accepted either, since other menus don't use them. Opinions welcomed.
NOTE: BLI_winstuff.h was meant to be a wrapper around windows.h to handle
undefining various crap that windows.h defines. Platform specific headers
should only have to be included in a few places. This reduces the number
of inclusions of BLI_winstuff.h to 16 which is a much more reasonable
number (than the 144 or whatever it used to be)
-- Ton caught a redraw trying to free a busy slider button in a script. Made sure the redraws don't happen in the scripts win during slider updates, but the realtime slider (automatically updates the 3d view) isn't working yet as it probably shouldn't with the current code. There's a pre 2.25 comment in Draw.c about it being hackish and wrong;
-- Melchior FRANZ reported a bug related to a script dict being freed too early (when two instances of the same script were on: one with a gui and the other with a file selector -- selecting the file triggered the problem);
-- bug #2074 reported by Roland Hess: totally empty string button -- no prepended but->str (e.g. "OB:") neither but->poin (the actual string) -- crashing Blender. Crash happened on editscreen.c: screen_delayed_push_undo calls strncpy() without checking the passed string. Didn't mess with it, better let the coder in charge of undo (Ton, right?) choose a preferred way to deal with it (report error, simply use blank string). Staying on the python side at least for now, I just made but->str = " " (a space) when user doesn't define it, so the crash shouldn't happen anymore.
http://projects.blender.org/tracker/?func=detail&atid=125&aid=2074&group_id=9
Thanks guys for reporting the problems.
- patch by Jonathan Merritt: new function Blender.Draw.Image() + its doc;
- patch by Campbell Barton: .users attribute for many types;
- tiny internal change in nmesh.hasVertexUV() method.
Thanks to Jonathan and Campbell for the patches and the patience!
- Small doc update in a script;
- Fixed bug #1742:
http://projects.blender.org/tracker/?func=detail&atid=125&aid=1742&group_id=9
It was an internal error in bpython. I was using G.main->script.last to find the currently running (if any) script, but this isn't reliable, we must check each open script to find out if one of them has the SCRIPT_RUNNING bitflag set.
Thanks intrr for reporting and blendix for pointing how to reproduce the bug. From my tests it should be working fine now.
Run everything thru indent to cleanup spaces vs tabs.
Clean up some of the comments by hand.
BGL.c was not touched due to all that macro wackyness.
There are no functional changes to the code.
Pre-indent versions of source are tagged with
tag bpy-cleanup-20040925 , just in case.
- hotkeys, obdatacopier and renameobjectbyblock, all from Jean-Michel Soler (jms);
- bevel_center by Loic Berthe, suggested for inclusion by jms;
- doc_browser, by Daniel Dunbar (Zr)
Thanks to them for the new contributions!
(I included doc_browser at 'Misc' because only users interested in script writing would actually use it, but it could also be under 'Help'. Opinions?)
BPython related:
- Added scriptlink methods to object, lamp, camera and world.
- Object: added object.makeTrack and object.clearTrack (old track method).
- sys: made sys.exists(path) return 0 for not found; 1 for file, 2 for dir and -1 for neither.
- doc updates and fixes.
- made ONLOAD event work. G.f's SCENESCRIPT bit was being zeroed in set_app_data.
- Blender: updated functions Load and Save to support the builtin importers and exporters besides .blend (dxf, videoscape, vrml 1.0, stl, ...)
- Draw: added mouse wheel events.
- Scene: added scene.play to play back animations (like ALT+A and SHIFT+ALT+A). Makes a good counter, too, when the 'win' attribute is set to a space that doesn't "animate".
The scene.play() addition and the fix to ONLOAD scriptlinks is part of the work for a Blender demo mode. It already works, but I'll still add support for Radiosity calculations and fix a thing in main(): it executes onload scripts too early (BIF_Init), giving funny results in alt+a animations and renderings when firing up Blender. Loading after the program is up has no such problems. When I finish I'll post examples of demo mode scripts.
- Added KlopUtils by Carlos Lopez and Batch Object Name Edit by Campbell Barton (thanks both!)
BPython:
- Fixed things I recently broke in Blender.sys, added Blender.sys.join(dir, file)
- doc and other minor updates
PupStrInput, a wrapper for the Blender String popup (thanks!)
- Fixed bug #1374 reported by Gabriel Beloin (gabio, thanks too):
http://projects.blender.org/tracker/?func=detail&atid=125&aid=1374&group_id=9
There was a minor mistake in the import menu: vrml called dxf and vice-versa and shortcuts were wrong (removed them).
- Doc updates, minor updates elsewhere.
- Campbell Barton contributed another function (thanks again), the Image.reload method:
with this a script can keep an image that is being edited and saved by an external program updated in Blender.
- updated docs with recently added functions
- dynamic menu entries now are sorted (alphabetically, of course)
- added new menu categories for scripts: Wizards, Modifiers, Generators, Materials, Animation:
only added to list of options, didn't mess with any Blender header. They are already available from the "Scripts" menu in the scripts win, but not elsewhere.
- added option 'datadir' to Blender.Get(option):
so scripts can use .blender/bpydata for reading / writing their data files.
-- Added two new functions to the Draw module, to display int and float popup buttons:
Both contributed by Campbell Barton, thanks!
-- Fixed a couple warnings.
(excuse me for doing all in a single commit, but they are tiny
fixes and it's bpython, that dark corner ...)
#1025 - FileSelector SEGV on dynamic callback Category:
Can't reproduce with current cvs, I'd say recent changes to fix
another crash related to FileSelector in gui-less scripts solved this
one, too.
#1028 - Reserved button event number:
Menu choices generate two events, one extra related to the menu
itself, with value=4. Made bpython ignore this extra event.
#1068 - FileSelector No file extension support:
As Ton wrote there, Blender itself doesn't support this yet. But the
requester also wanted Window.File/ImageSelector to accept a pathname. Done. Also updated doc.
#959 - Segfault on background rendering:
This happened in bg mode (blender -b filename -a, for example) when
a script with the line "Blender.Redraw()" was linked to FRAMECHANGED events. As reported in the bug page, it was because curarea is NULL in bg mode. Made Redraw() check for this and not call functions that expected curarea in Redraw, like one to swap buffers.
#1072 - Blender.Redraw() Segfault:
Good catch : ). Scripts called from the scripts win that called
Blender.Redraw() or Blender.Window.Redraw() would crash Blender because of a dirty pointer in Spacescript->script. Fixed.
- tentative fix for scripts with CR/LF endings and split lines:
in 2.32, the ac3d and vrml2 exporters, for example, had lines
split with '\\\\' and so gave syntax errors when executed on Win
platforms, because the scripts bundled with Win binaries had dos
line endings.
- Chris Keith has written code to execute Python scripts from the
command-line, with '-P ' switch: "blender -P filename":
a Blender.Quit function was also added, so Blender can quit after
running the script (end the script with Blender.Quit()), but there's
still work to be done in this part, including adding more functions,
to load / save .blend files and to run scripts. More testing and
discussions are necessary.
Thanks Chris, for both your contributions and your patience, since I
wasn't available to check / commit this for a while.
data definitions from .h files into corresponding .c files.
Blame zr for this since he's the one who pointed out that our
bpy headers were a mish-mash of stuff that belonged in the .c files!
In a nutshell, the headers should contain the declarations necessary
to use a module or class. The implementation files ( .c in this case )
should contain statements that allocate storage ( definitions in
the C sense ) and executable code.
When used at file scope, the keyword 'static' means "don't tell
anyone else about this". Since headers describe a public
interface, static declarations and definitions belong in the
implementation files.
The net result of all this is that after stuff has moved out
into the .c files, the .h files are empty or mostly empty.
I didn't delete them since there seem to be some public
declarations and because I did not want to cause too much
disruption at one time. Time enough for that later!
-- removed struct Script (DNA_script_types.h) from makesdna to blender/include/BPI_script.h
(BPI meaning Blender Python-related external Include file).
Had agreed with Ton that makesdna was not the proper place for it.
-- fixed two small warnings in Ipo.c (variables might be used uninitialized)
-- fixed a bug reported on blender.org's python forum by Wim Van Hoydonck (aka tuinbels):
Blender would hang if a script failed. My fault, accidentally put a node=node->next type call outside the while loop check, so it never ended.
With makesdna/DNA_script_types.h removed and include/BPI_script.h added, msvc projectfiles will need to be updated. Sorry to do it now, but I promissed I'd fix this before next release.
- add a new space: Space Script
- add a new dna struct: Script
- add these two properly everywhere they are meant to
It's not a tiny commit, but most of it is ground work for what is still to be done.
Right now the benefits should be: freeing the Text Editor to be used in a window even while a script w/ gui in "on" and letting more than one currently running script w/ gui be accessible from each window
Some files are added, so some build systems (not autotools) will need updates
- Object: implemented getBoundBox and makeDisplayList methods
- NMesh and Object: small internal changes for nicer behavior
- Draw: added function PupMenu
- Docs: updated for the additions above
Auto build tiny fix: added the imbuf include dir to source/creator/Makefile.am
there)
* replaced with nice local (uiBlock) storage of what goes on in drawing.
it now only calls a glDrawBuffer() and glFinish() when it is actually
needed
Result: interface drawing in general is speedy again, especially for gfx
cards that dont allow frontbuffer drawing, and copy stuff to the frontbuf
with a glFinish() call.
Needs to be tested on all platforms... report to me when you see problems
like menus not drawing correctly, tooltips not drawing or not disappearing,
etc.
Thanks Florian Eggenberger for telling us about it. And Greg McBride for
pointing a possible fix.
- Draw.Text and Draw.GetStringWidth updated:
Now they accept an optional second parameter to set font size and Draw.Text
returns the drawn string width.
- Partially fixed the update() and PutRaw() bugs in NMesh:
A total fix will require bigger changes, but what was done (unless buggy)
takes care of the common cases.
Added a function call to creator.c that is needed by exppython, but forgot
to add the function also to the old bpython implementation. Thanks, Hos!
- Addition in Draw.Text and Draw.GetStringWidth (Python Draw methods):
Now script writers can select the font size: normal, small or tiny.
