summaryrefslogtreecommitdiff
path: root/lib
AgeCommit message (Collapse)Author
2013-09-12gras: added thread pool to global configJosh Blum
2013-09-09gras: work on global config that works on hierJosh Blum
2013-09-05gras: move sbuffer inlines into libJosh Blum
This avoids inline ASM to be LLVM JIT friendly. This commit includes a unit test for jit factory + sbuffer.
2013-08-30gras: use Theron's Utils.h for numa allocJosh Blum
2013-08-30gras: fix jit factory declaration when no LLVMJosh Blum
2013-08-06gras: missing ifdef for jit factory w/o llvmJosh Blum
2013-08-05gras: removed extra slashesJosh Blum
2013-08-05gras: jit factory use FindProgramByNameJosh Blum
2013-08-05gras: work on using clang api - but commented outJosh Blum
The last thing missing is that the API use case doesnt automatically have default C++ include paths.
2013-08-04gras: jit factory automatically adds GRAS_ROOT include pathJosh Blum
2013-08-01gras: jit factory api + unit testsJosh Blum
2013-07-31gras: take care of some misc warningsJosh Blum
2013-07-30gras: rename some factory functionsJosh Blum
2013-07-28gras: fixed typo on exceptionsJosh Blum
2013-07-27gras: just warn if the factory is already registeredJosh Blum
2013-07-27gras: added arg check with nice warning to callable and factoryJosh Blum
2013-07-27gras: work on module loading and python factoryJosh Blum
naming convention, path convention, env var convention, And, make a python factory -- but its a little halfassed. You see, we didnt do the swig director thing yet, so C++ cant yet create python blocks from the factory.
2013-07-25gras: make ptree usage optional on old boostsJosh Blum
2013-07-23gras: minor fix for closed feedback loopJosh Blum
This updates the input available state after handling preload. Thanks Devin!
2013-07-17gras: module loading is recursive and takes filesJosh Blum
2013-07-15gras: make all start/commit messages high-prioJosh Blum
This makes doing live topology changes w/ commit instant, even if much buffering is backed up.
2013-07-14gras: fixed allocation override bug + unit testJosh Blum
When multiple consumers all choose to override the buffer, they were doing so with the same token, this lead to some buffer allocation confusion and nasty errors. Its ok for one to win out, but each allocator needs a unique token. There is also a unit test to check for this behaviour.
2013-07-14gras: circ buff mutex for dealloc as wellJosh Blum
2013-07-13gras: always use ns for TimeTagJosh Blum
The msvc performance counter tps isnt very high. And as it turns out, we dont need to care, since its all internal to the implementation.
2013-07-10gras: ptree to pmc try max width fixed intJosh Blum
long long has pretty bad swig wrapping, its easier to use fixed width int64s for now.
2013-07-09gras: default to widest integer converterJosh Blum
2013-07-09gras: rename props files to calls (for callable work)Josh Blum
2013-07-07gras: added module loaderJosh Blum
2013-07-07gras: python hooks for factoryJosh Blum
2013-07-07gras: created element factoryJosh Blum
2013-07-07gras: renamed key to name for clarityJosh Blum
2013-07-07gras: restore property access w/ jsonJosh Blum
2013-07-06gras: locate_element since now elements have callsJosh Blum
2013-07-05gras: pythonic work on callable interfaceJosh Blum
2013-07-05gras: begin replacing set/get with callableJosh Blum
2013-07-05gras: make callable a copyable classJosh Blum
2013-07-05gras: serious template insanityJosh Blum
2013-07-05gras: save callable work compiling w/ minor unit testJosh Blum
2013-07-05gras: revert block changes, test w/ callable nowJosh Blum
2013-07-04gras: function registry is more flexibleJosh Blum
2013-07-03gras: use the base class settings calls to cleanup codeJosh Blum
2013-06-23gras: PMC and serialization hooks for Tag typeJosh Blum
2013-06-22gras: various fixes for new time tag classJosh Blum
2013-06-22gras: use inline queue for post output bufferJosh Blum
* simplifies logic * fixes bug #103 An assertion fires if multiple post output buffers are used per work(). This could be fixed in the future if its truely needed.
2013-06-21gras: adding pythonic interface to time tagJosh Blum
2013-06-21gras: first cut at time tag classJosh Blum
2013-06-20gras: work on goddamn done logicJosh Blum
How blocks mark themselves done has been one of the most annoying things in this development. This done logic is only valuable for QA tests, it doesnt even exist in the practical use case. How it works now: * blocks mark done when sync inputs are done or all inputs are done * removed the force_done input config, its no longer needed * the wait() implementation gives blocks a grace period between an input becoming done and the block itself becoming done. After the grace period, the block is forced done.
2013-06-20gras: moved all done logic end of task mainJosh Blum
2013-06-19gras: add retries for make_circular_buffer ipc exceptionsJosh Blum
This commit addresses #104
2013-06-19gras: query_topology visualize done portsJosh Blum
generated by cgit (git 2.34.1) at 2025-03-13 09:59:57 +0000