The goal is to make the search faster to use by dynamically adapting to the user.
This can be achieved using the simple but common approach of showing recently
selected items at the top. Note, that the "matching score" between the query and
each search item still has precedence when determining the order. So the last used
item is only at the top, if there is no other search item that matches the query better.
Besides making the search generally faster to use, my hope is that this can also
reduce the need for manually weighting search items in some places. This is
because while the ordering might not be perfect the first time, it will always be
once the user selected the element that should be at the top once.
This patch includes:
* Support for taking recent searches into account in string searching.
* Keep track of a global list of recent searches.
* Store recent searches on disk similar to recently opened files.
* A new setting in the user preferences that allows disabling the functionality.
This can be used if deterministic key strokes are required, e.g. for automated tests.
In the future this could be improved in different ways:
* Add some kind of separator in the search list to indicate which elements are at
the top because they have been used recently.
* Store the recent search items per search, instead of in a global list. This way
it could adapt to the user even better.
Pull Request: https://projects.blender.org/blender/blender/pulls/110828
Ensure the string isn't null terminated before len, if this happens
it's likely an error calculating the length argument. Since the length
is passed to memcpy it could result in reading outside `str` bounds.
There was one case where the string was duplicated then grew by 1 byte
afterwards. Replace this with an allocation since it's an uncommon
operation, not especially useful to support.
Make type conversions explicit so it's clear when char/char32_t/uint
values are being mixed, also use int instead of size_t for cursor
functions because the cursor is an int - which caused many int/size_t
comparisons.
This makes it clearer other "safe" functions should be used in
combination with the resulting offsets.
Also correct doc-string which wasn't updated from the "or_error()"
version of this function.
The functions used to calculate the UTF8 code-points already used the
safe versions (count_utf8_code_points & BLI_str_utf8_as_unicode_step).
So it makes sense to use safe accessors elsewhere too.
There were enough cases of callers ignoring a potential the error value,
using the column width for e.g. to calculate pixel sizes, or the size in
bytes to calculate buffer offsets.
Since text fields & labels can include characters that return an error
from BLI_str_utf8_as_unicode, add the suffix to make this explicit.
Strings that include Latin1 encoding or corrupt UTF8 byte sequences
could read past the buffer bounds (stepping over the null terminator).
Resolve by passing in the string length.
Other changes to support non-UTF8 byte sequences:
- BLI_str_utf8_offset_{to/from}_index were accumulating
the UTF8 offset without accounting for non-UTF8 characters
which could cause a buffer underflow or enter an eternal loop.
- BLI_str_utf8_offset_to_index would read past the buffer bounds if the
offset passed in if it was in the middle of a UTF8 byte sequence.
BLI_str_utf8_char_width returns -1 for character without a known width.
This caused the right cursor motion to skip these characters.
While editable text should not contain control characters,
cursor motion should behave properly in cases when they do.
Since hiding symbols on Linux, in many cases only addresses are printed.
This utility can run run on the back-trace to replace addresses
with line & function information.
See: ./tools/utils/addr2line_backtrace.py --help for usage information.
Note that some examples online run addr2line directly and use the output
in the stack-trace, while convenient and acceptable in some cases, in my
tests addr2line can take over 20 seconds to complete for a single
address. Implement this as a post-process instead. Multi-processing to
prevent this taking too long (around ~23 seconds on my system).
Ref !111416.
This fixes#111767
`In Front` objects remain visible even if they are behind non
`In Front` objects.
It is to be expected then that the snap for them is not occluded as if
they were not `In Front`.
The f12e9f32b5 patch introduced a new improved method of blending
dual quaternion transformations to handle combined scale and rotation
better. However, the changes were not complete:
* The new math ignored crazyspace computations, which need to compute
a complete transformation matrix. As an optimization, the new method
avoided fully computing the scale component, so the matrix would
have no scale or shear.
* The Armature constraint is supposed to behave identically to the
modifier, and it was not updated. The constraint also requires
computing a complete matrix.
This change extracts the new math into a utility function, change
the optimization to be controlled by a parameter, and use the new
function in the constraint.
Pull Request: https://projects.blender.org/blender/blender/pulls/111759
The hash tables and vector blenlib headers were pulling many more
headers than they actually need, including the C base math header,
our C string API header, and the StringRef header. All of this
potentially slows down compilation and polutes autocomplete
with unrelated information.
Also remove the `ListBase` constructor for `Vector`. It wasn't used
much, and making it easy to use `ListBase` isn't worth it for the
same reasons mentioned above.
It turns out a lot of files depended on indirect includes of
`BLI_string.h` and `BLI_listbase.h`, so those are fixed here.
Pull Request: https://projects.blender.org/blender/blender/pulls/111801
The `EdgeHash` and `EdgeSet` data structures are designed specifically
as a hash of an order agnostic pair of integers. This specialization can
be achieved much more easily with the templated C++ data structures,
which gives improved performance, readability, and type safety.
This PR removes the older data structures and replaces their use with
`Map`, `Set`, or `VectorSet` depending on the situation. The changes
are mostly straightforward, but there are a few places where the old
API made the goals of the code confusing.
The last time these removed data structures were significantly changed,
they were already moving closer to the implementation of the newer
C++ data structures (aa63a87d37).
Pull Request: https://projects.blender.org/blender/blender/pulls/111391
Including <iostream> or similar headers is quite expensive, since it
also pulls in things like <locale> and so on. In many BLI headers,
iostreams are only used to implement some sort of "debug print",
or an operator<< for ostream.
Change some of the commonly used places to instead include <iosfwd>,
which is the standard way of forward-declaring iostreams related
classes, and move the actual debug-print / operator<< implementations
into .cc files.
This is not done for templated classes though (it would be possible
to provide explicit operator<< instantiations somewhere in the
source file, but that would lead to hard-to-figure-out linker error
whenever someone would add a different template type). There, where
possible, I changed from full <iostream> include to only the needed
<ostream> part.
For Span<T>, I just removed print_as_lines since it's not used by
anything. It could be moved into a .cc file using a similar approach
as above if needed.
Doing full blender build changes include counts this way:
- <iostream> 1986 -> 978
- <sstream> 2880 -> 925
It does not affect the total build time much though, mostly because
towards the end of it there's just several CPU cores finishing
compiling OpenVDB related source files.
Pull Request: https://projects.blender.org/blender/blender/pulls/111046
Fix an UBSAN warning about undefined behavior, by preventing additions
to `nullptr`:
```
blender/source/blender/blenlib/intern/BLI_memiter.c:136:42: runtime error:
applying non-zero offset 48 to null pointer
```
No functional changes.
This PR adds the Lacunarity and Normalize inputs to the Noise node
similar to the Voronoi node.
The Lacunarity input controls the scale factor by which each
successive Perlin noise octave is scaled. Which was previously hard
coded to a factor of 2.
The Noise node normalizes its output to the [0, 1] range by default.
The Normalize option makes it possible for the user to disable that.
To keep the behavior consistent with past versions it is enabled by
default.
To make the aforementioned normalization control easer to implement,
the fractal noise code now accumulates signed noise and remaps the
final sum, as opposed to accumulating positive [0, 1] noise.
Pull Request: https://projects.blender.org/blender/blender/pulls/110839
Listing the "Blender Foundation" as copyright holder implied the Blender
Foundation holds copyright to files which may include work from many
developers.
While keeping copyright on headers makes sense for isolated libraries,
Blender's own code may be refactored or moved between files in a way
that makes the per file copyright holders less meaningful.
Copyright references to the "Blender Foundation" have been replaced with
"Blender Authors", with the exception of `./extern/` since these this
contains libraries which are more isolated, any changed to license
headers there can be handled on a case-by-case basis.
Some directories in `./intern/` have also been excluded:
- `./intern/cycles/` it's own `AUTHORS` file is planned.
- `./intern/opensubdiv/`.
An "AUTHORS" file has been added, using the chromium projects authors
file as a template.
Design task: #110784
Ref !110783.
Using ClangBuildAnalyzer on the whole Blender build, it was pointing
out that BLI_math.h is the heaviest "header hub" (i.e. non tiny file
that is included a lot).
However, there's very little (actually zero) source files in Blender
that need "all the math" (base, colors, vectors, matrices,
quaternions, intersection, interpolation, statistics, solvers and
time). A common use case is source files needing just vectors, or
just vectors & matrices, or just colors etc. Actually, 181 files
were including the whole math thing without needing it at all.
This change removes BLI_math.h completely, and instead in all the
places that need it, includes BLI_math_vector.h or BLI_math_color.h
and so on.
Change from that:
- BLI_math_color.h was included 1399 times -> now 408 (took 114.0sec
to parse -> now 36.3sec)
- BLI_simd.h 1403 -> 418 (109.7sec -> 34.9sec).
Full rebuild of Blender (Apple M1, Xcode, RelWithDebInfo) is not
affected much (342sec -> 334sec). Most of benefit would be when
someone's changing BLI_simd.h or BLI_math_color.h or similar files,
that now there's 3x fewer files result in a recompile.
Pull Request #110944
In RNA collections storing ID references, the name of the collection
item may not always be unique, when several IDs from different libraries
are present.
While rare, this situation can become deadly to liboverride, by causing
random but exponential liboverride hierarchies corruptions.
This has already been alleviated by using preferably both name and index
in items lookup (a05419f18b) and by reducing the risk of name collision
in general between liboverrides and their linked reference (b9becc47de).
This commit goes further, by ensuring that references to items of RNA
collections of IDs stored in liboverride operations become completely
unambiguous. This is achieved by storing an extra pointer to the item's
ID itself, when relevant.
Lookup then requires a complete match `name + ID` to be successful,
which is guaranteed to match at most a single item in the whole RNA
collection (since RNA collection of IDs do not allow duplicates, and
the ID pointer is always unique).
Note that this ID pointer is implemented as an `std::optional` one
(either directly in C++ code, or using an new liboverride operation `flag`
in DNA). This allows to smoothly transition from existing data to the
added ID pointer info (when needed), without needing any dedicated
versioning. This solution also preserves forward compatibility as much
as possible.
It may also provide marginal performances improvements in some cases, as
looking up for ID items in RNA collections will first check for the
ID pointer, which should be faster than a string comparision.
Implements #110421.
Pull Request: https://projects.blender.org/blender/blender/pulls/110773
Part 1/3 of #109135, #110272
Adds a new DNA structure for defining node group interfaces without
using `bNodeSocket` and with additional node UI item types.
Node group interfaces are organized as a hierarchy of "items", which
can be sockets or panels. Panels can contain both sockets and other
panels (although nested panels beyond the root panel may be disabled to
avoid complexity on the user level).
Sockets can be added to the interface in any order, not just the
conventional outputs..inputs order. Sockets can be marked as both input
and output, generating 2 sockets on node instances.
The C++ API in the DNA struct allows manipulating the interface
declaration by adding and removing items, moving them inside the
interface or into a different panel.
Pull Request: https://projects.blender.org/blender/blender/pulls/110885
Both the `Math` node and the `Vector Math` currently only explicitly
support modulo using truncated division which is oftentimes not the
type of modulo desired as it behaves differently for negative numbers
and positive numbers.
Floored Modulo can be created by either using the `Wrap` operation or
a combination of multiple `Math` nodes. However both methods obfuscate
the actual intend of the artist and the math operation that is actually
used.
This patch adds modulo using floored division to the scalar `Math` node,
explicitly stating the intended math operation and renames the already
existing `"Modulo"` operation to `"Truncated Modulo"` to avoid confusion.
Only the ui name is changed, so this should not break compatibility.
Pull Request: https://projects.blender.org/blender/blender/pulls/110728
The root cause was a classic fixed-size epsilon issue. The code that
checked if an fcurve was effectively flat, and thus shouldn't be
normalized, used a fixed-size epsilon that was reasonable for values
close-ish to zero, but didn't work well for values >= 1.0.
This patch addresses the issue by introducing a new function
`ulp_diff_ff()` that robustly computes the number of floating point
steps between two floats, and using that to ensure that a minimum
number of representable floats exist between the min/max values
of the curve. This approach scales appropriately up and down to
both huge and tiny values.
This patch also updates the existing `compare_ff_relative()` function
to use the new robust ulps code for the ulps-based part of its
comparison, resolving an issue documented in its unit tests where
it behaved poorly for values close to zero.
Pull Request: https://projects.blender.org/blender/blender/pulls/110796
Implements part of #101689.
The "poly" name was chosen to distinguish the `MLoop` + `MPoly`
combination from the `MFace` struct it replaced. Those two structures
persisted together for a long time, but nowadays `MPoly` is gone, and
`MFace` is only used in some legacy code like the particle system.
To avoid unnecessarily using a different term, increase consistency
with the UI and with BMesh, and generally make code a bit easier to
read, this commit replaces the `poly` term with `poly`. Most variables
that use the term are renamed too. `Mesh.totface` and `Mesh.fdata` now
have a `_legacy` suffix to reduce confusion. In a next step, `pdata`
can be renamed to `face_data` as well.
Pull Request: https://projects.blender.org/blender/blender/pulls/109819
