CVS log for pkgsrc/devel/hyperscan/Makefile

Up to [cvs.NetBSD.org] / pkgsrc / devel / hyperscan

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

hyperscan: updated to 5.4.2

5.4.2

Roll back bugfix for github issue 350: Besides using scratch for
corresponding database, Hyperscan also allows user to use larger scratch
allocated for another database. Users can leverage this property to achieve
safe scratch usage in multi-database scenarios. Behaviors beyond these are
discouraged and results are undefined.
Fix hsdump issue due to invalid nfa type.


5.4.1

The Intel Hyperscan team is pleased to provide a bug fix release to our open source library.
Intel also maintains an upgraded version available through your Intel sales representative.
Bugfix for issue 184: fix random char value of UTF-8.
Bugfix for issue 291: bypass logical combination flag in hs_expression_info().
Bugfix for issue 292: fix build error due to libc symbol parsing.
Bugfix for issue 302/304: add empty string check for pure literal API.
Bugfix for issue 303: fix unknown instruction error in pure literal API.
Bugfix for issue 303: avoid memory leak in stream close stage.
Bugfix for issue 305: fix assertion failure in DFA construction.
Bugfix for issue 317: fix aligned allocator segment faults.
Bugfix for issue 350: add quick validity check for scratch.
Bugfix for issue 359: fix glibc-2.34 stack size issue.
Bugfix for issue 360: fix SKIP flag issue in chimera.
Bugfix for issue 362: fix one cotec check corner issue in UTF-8 validation.
Fix other compile issues.


5.4.0

Improvement on literal matcher "Fat Teddy" performance, including support for Intel(R) AVX-512 Vector Byte Manipulation Instructions (Intel(R) AVX-512 VBMI).
Introduce a new 32-state shuffle-based DFA engine ("Sheng32"). This improves scanning performance by leveraging AVX-512 VBMI.
Introduce a new 64-state shuffle-based DFA engine ("Sheng64"). This improves scanning performance by leveraging AVX-512 VBMI.
Introduce a new shuffle-based hybrid DFA engine ("McSheng64"). This improves scanning performance by leveraging AVX-512 VBMI.
Improvement on exceptional state handling performance for LimEx NFA, including support for AVX-512 VBMI.
Improvement on lookaround performance with new models, including support for AVX-512.
Improvement on DFA state space efficiency.
Optimization on decision of NFA/DFA generation.
hsbench: add CSV dump support for hsbench.
Bugfix for cmake error on Icelake under release mode.
Bugfix in find_vertices_in_cycles() to avoid self-loop checking in SCC.
Bugfix for issue 270: fix return value handling in chimera.
Bugfix for issue 284: use correct free function in logical combination.
Add BUILD_EXAMPLES cmake option to enable example code compilation.
Some typo fixing.

hyperscan: set LICENSE

*: recursive bump for icu 76 shlib major version bump

*: revbump for icu downgrade

*: recursive bump for icu 76.1 shlib bump

*: replace CMAKE_ARGS with CMAKE_CONFIGURE_ARGS

revbump after icu and protobuf updates

hyperscan: update HOMEPAGE/MASTER_SITES

*: recursive bump for icu 74.1

*: recursive bump for Python 3.11 as new default

devel: Adapt packages to use USE_(CC|CXX)_FEATURES

Mass-change BUILD_DEPENDS to TOOL_DEPENDS outside mk/.

Almost all uses, if not all of them, are wrong, according to the
semantics of BUILD_DEPENDS (packages built for target available for
use _by_ tools at build-time) and TOOL_DEPEPNDS (packages built for
host available for use _as_ tools at build-time).

No change to BUILD_DEPENDS as used correctly inside buildlink3.

As proposed on tech-pkg:
https://mail-index.netbsd.org/tech-pkg/2023/06/03/msg027632.html

*: GCC_REQD must always be appended to.

revbump after textproc/icu update

massive revision bump after textproc/icu update

revbump for textproc/icu update

revbump for icu and libffi

revbump for textproc/icu

hyperscan: Handle sqlite dependency - attempt to fix build

PKGREVISION bump for anything using python without a PYPKGPREFIX.

This is a semi-manual PKGREVISION bump.

devel/hyperscan: Update to 5.0.0

New features:

- A hybrid library of Hyperscan and PCRE called Chimera
- Logical combination matching
- Experimental support for Windows

NetBSD has popcount, use it. Don't install non-portable examples.
Bump revision.

devel/hyperscan: Import version 4.7.0

Hyperscan is a high-performance multiple regex matching library. It
follows the regular expression syntax of the commonly-used libpcre
library, but is a standalone library with its own C API. Hyperscan
uses hybrid automata techniques to allow simultaneous matching of
large numbers (up to tens of thousands) of regular expressions and for
the matching of regular expressions across streams of data.