The unused result was reported by Clang-Tidy 11.
It does make sense to check for the failed mmap() calls rather than
quietly suppress errors.
In this change failures are reported, but application execution is
not aborted. This is a bit disputable, but it feels to be a safer
thing to do now.
It is unclear how to test the code though, as we don't have any
tools in-place to simulate read errors.
Differential Revision: https://developer.blender.org/D10223
BLI_strncpy_wchar_from_utf8 made the assumption that
wchar_t is UTF-32 bit regardless of environment, while
this holds true on both mac and linux, on windows
wchar_t is actually actually UTF-16.
This resulted in the upper 16 bits being dropped from
from some string conversions and prevented blender
from starting when installed in a path with unicode
code-points over 0xffff.
There was also a fair bit of code duplication between
BLI_strncpy_wchar_from_utf8 and BLI_str_utf8_as_unicode_and_size
this change essentially removes all logic from
BLI_strncpy_wchar_from_utf8 and calls the right function
for the right environment.
Reviewed By: brecht , Robert Guetzkow
Differential Revision: https://developer.blender.org/D9822
Important lesson to be learned here, leaving comments
is great and in the moment, they usually make sense.
Many months later they may not quite make as much sense
any more and time will have to be spend to figure out
what was meant, all of this would have been averted with a
better comment.
The zero terminator in this case, I can find no evidence
of it being used or relied on at any point. It does however
break GTests's `EXPECT_EXIT` macro that stops looking in
the output as soon as it sees the zts and doesn't end up
looking at the actual assert text being thrown. Which in
turn makes the`fcurve_active_keyframe`test fail when run
in debug mode on windows.
* tbb::blocked_range moved to a different namespace and since the fix is
non-trivial, remove some unused code that used this.
* Task group priorities are no longer supported. It's unclear if they are
useful at all right now and even set correctly, for now all tasks are equal
priority with TBB 2021.
Instead of submitting tons of tiny IO syscalls, we can speed things up
significantly by `mmap`ing the .blend file into virtual memory and directly
accessing it.
In my local testing, this speeds up loading the Dweebs file with all its
linked files from 19sec to 10sec (on Linux).
As far as I can see, this should be supported on Linux, OSX and BSD.
For Windows, a second code path uses `CreateFileMapping` and
`MapViewOfFile` to achieve the same result.
Reviewed By: mont29, brecht
Differential Revision: https://developer.blender.org/D8246
Use approximate geodesic distance computatiom that crosses through triangles
rather than only along edges. Using only edges would give artifacts already
on a simple grid.
Fixes T78752, T35590, T43393, T53602
Differential Revision: https://developer.blender.org/D10068
This patch does two things:
* Introduce a Seed to the random distribution method
* Bring in a new distribution method for the point scattering node
Patch Review: https://developer.blender.org/D9787
Note: This commit doesn't not handle doversion. Which means that users
need to manually update their files that were using the Point Distribute
node and reconnect inputs to the "Maximum Density" socket.
Original patch by Sebastian Parborg, with changes to not rely on the cy
libraries and overall cleanup.
Patch review by Jacques Lucke, besides help with the new "heap" system
that was required for this algorithm.
Based on Cem Yuksel. 2015. Sample Elimination for Generating Poisson Disk
Sample. Sets. Computer Graphics Forum 34, 2 (May 2015), 25-32
http://www.cemyuksel.com/research/sampleelimination/
It turns out that after the fix to T83196 (rB814b2787cadd) the matrix
to quaternion conversion can produce noncanonical results in large
areas of the rotation space, when previously this was limited to
way smaller areas. This in turn causes Swing+Twist math to produce
angles beyond 180 degrees, e.g. outputting a -120..240 range.
This fixes both issues, ensuring that conversion outputs a canonical
result, and decomposition canonifies its input.
This was reported in chat by @jpbouza.
BLI_getenv has always incorrectly returned ascii rather
than UTF-8. This change corrects this behaviour.
This resolves issues when the `BLENDER_USER_CONFIG`
environment variable contains a path with Unicode characters
on windows as reported in T74510 (but unlikely the root
cause for the issue at hand there)
Differential Revision: https://developer.blender.org/D9831
Reviewed by: brecht
I thought I had reasoned that the add_patch would only happen
when the patch was not already in a cell, but I missed reasoning
about merged cells. So switched to a set 'add' instead of 'add_new'.
This was returning an empty allocated string, however almost
all callers checked if the return value was NULL before freeing,
making for misunderstandings on the intended use of this function.
BCAnimationSampler::initialize_curves for example detected the
f-curves animation type to 'bone' based on a non-NULL return value
which never failed.
Also fixes two leaks where the the result of BLI_str_quoted_substrN
wasn't freed.
BLI_str_quoted_substrN didn't unescape the resulting string,
yet all callers were using this with animation paths which are
created using BLI_str_escape().
- Fix BLI_str_quoted_substrN use with escaped strings by calling
BLI_str_unescape().
Note that it's possible we want an a version of this function that
keeps escape characters. This could be added if it's required.
- Fix end quote detection using BLI_str_escape_find_quote
checking for `\"` isn't reliable since an even number of back-slashes
before a quote means it's not escaped.
Caused problems in the Asset Browser branch when passing an empty path. The
function shouldn't blindly add a slash but sanity-check the input parameters
first.
Replace `typedef` with `using` in C++ code.
In the case of `typedef struct SomeName { ... } SomeName;` I removed the
`typedef` altogether, as this is unnecessary in C++. Such cases have been
rewritten to `struct SomeName { ... };`
No functional changes.
This is the initial merge from the geometry-nodes branch.
Nodes:
* Attribute Math
* Boolean
* Edge Split
* Float Compare
* Object Info
* Point Distribute
* Point Instance
* Random Attribute
* Random Float
* Subdivision Surface
* Transform
* Triangulate
It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier.
Notes on the Generic attribute access API
The API adds an indirection for attribute access. That has the following benefits:
* Most code does not have to care about how an attribute is stored internally.
This is mainly necessary, because we have to deal with "legacy" attributes
such as vertex weights and attributes that are embedded into other structs
such as vertex positions.
* When reading from an attribute, we generally don't care what domain the
attribute is stored on. So we want to abstract away the interpolation that
that adapts attributes from one domain to another domain (this is not
actually implemented yet).
Other possible improvements for later iterations include:
* Actually implement interpolation between domains.
* Don't use inheritance for the different attribute types. A single class for read
access and one for write access might be enough, because we know all the ways
in which attributes are stored internally. We don't want more different internal
structures in the future. On the contrary, ideally we can consolidate the different
storage formats in the future to reduce the need for this indirection.
* Remove the need for heap allocations when creating attribute accessors.
It includes commits from:
* Dalai Felinto
* Hans Goudey
* Jacques Lucke
* Léo Depoix
Adjust the threshold for switching from the base case to trace > 0,
based on very similar example code from www.euclideanspace.com to
avoid float precision issues when trace is close to -1.
Also, remove conversions to and from double, because using double
here doesn't really have benefit, especially with the new threshold.
Finally, add quaternion-matrix-quaternion round trip tests with
full coverage for all 4 branches.
Differential Revision: https://developer.blender.org/D9675
By checking if a cell has already been processed in the finding patch
component code, an enormous speedup happens. This only will be
noticeable if there are lots of patches, and some cells with a
large number of patches.
In case where there are coplanar instersections where
each part has a lot of triangles, the finding-cells algorithm was
very inefficient. This uses a Set instead of a Vector to keep track
of a cell's patches, avoids going through all patch x patch combinations,
avoids going through all patches to renumber after a merge, and
merges smaller patch-sixe cells into larger ones.
All this reduces the time to find cells in the cited case by a factor of 10.
Since D6133 fluid particle code uses thread local storage to collect
springs created during a time step before adding them to the actual
spring array.
Prior to the switch to TBB there was a single finalize callback which
was called on the main thread, so it could use psys_sph_flush_springs
and insert the new entries into the final buffer. However in D7394 it
was replaced with a reduce callback, which is supposed to be thread
safe and have no side effects. This means that the only thing it can
safely do is copy entries to the other temporary buffer.
In addition, careful checking reveals that the 'classical' solver
doesn't actually add springs, so reduce isn't needed there.
Differential Revision: https://developer.blender.org/D9632
The existing hash function didn't work well with Set's method of
masking to the lower bits, because many verts have zeros in the
lower bits.
Also, replaced VectorSet with Set for Vert deduping.
By using floating point filters, the speed improves by a factor of 2 to 10.
This will help speed up some cases of the Exact Boolean modifier.
Changed the interface of mpq2::isect_seg_seg to not return mu, as it was
not needed and not calculating it saved 15% time.
