- Nov 14, 2016
-
-
Mick Jordan authored
-
Mick Jordan authored
-
Stepan Sindelar authored
Merge pull request #459 in G/fastr from ~STEPAN.SINDELAR_ORACLE.COM/fastr:jvmtrend-perf-fixes to master * commit '083469db': Explanatory comment in ClassHierarchyNode ValueProfiles in specials + few other places
-
- Nov 11, 2016
-
-
stepan authored
-
- Nov 10, 2016
-
-
Mick Jordan authored
-
Mick Jordan authored
-
Mick Jordan authored
Merge pull request #461 in G/fastr from ~MICK.JORDAN_ORACLE.COM/fastr:feature/internal-pkgtest to master * commit 'a2e61505': pkgtest: document —no-install option pkgtest: refactor to use internal GNU R when possible; misc other changes for daily recommended packages test
-
Mick Jordan authored
-
- Nov 09, 2016
-
-
Mick Jordan authored
pkgtest: refactor to use internal GNU R when possible; misc other changes for daily recommended packages test
-
Mick Jordan authored
Merge pull request #458 in G/fastr from ~MICK.JORDAN_ORACLE.COM/fastr:bugfix/solaris-build-2 to master * commit '59d07024': set XZ_HOME package download, fix -m64 regression in GNU R build for Solaris, solaris arch is sparcv9
-
stepan authored
-
Mick Jordan authored
set XZ_HOME package download, fix -m64 regression in GNU R build for Solaris, solaris arch is sparcv9
-
- Nov 08, 2016
-
-
Lukas Stadler authored
* commit '91be8650': small cleanups fix to argument matching: don't complain if args like (xy, x) are matched exactly implement Rf_classgets small fixes in native code fix imports refactor isVector (cache mode to type conversion)
-
Mick Jordan authored
* commit '9659c84f': changes for R-3.3.1
-
- Nov 07, 2016
-
-
Mick Jordan authored
-
Lukas Stadler authored
-
Lukas Stadler authored
-
Stepan Sindelar authored
Merge pull request #422 in G/fastr from ~STEPAN.SINDELAR_ORACLE.COM/fastr:feature/replace-refactoring to master * commit '3a5fb0c7': ReplacementBlockNode -> ReplacementDispatchNode Create special replacement even for <<- RCodeBuilder has context with temp var names start index Additional list extract tests Replacement executes the target first to specialize Replacement refactoring
-
Lukas Stadler authored
-
Lukas Stadler authored
-
Lukas Stadler authored
-
Lukas Stadler authored
-
Lukas Stadler authored
* commit 'ac6d0bb8': install order matters for recommended packages
-
- Nov 05, 2016
-
-
Mick Jordan authored
-
- Nov 04, 2016
-
-
Mick Jordan authored
* commit '6031acb5': mx.fastr: remove nocompile in _set_graal_options remove redundant mx rcmplib support jars and dirs in DefaultResourceHandlerfactory add explicit method for accessing directory "resources" mx_fastr_dists: add doc comment properly ensure that recommended packages can be built in all situations and are included in the release
-
Mick Jordan authored
-
stepan authored
plus remove unnecessary try-finally and obsolete TODO comment
-
stepan authored
Now when the reading and writing of the target is responsibility of ReplacementNode, SpecialReplacementNode works even with <<- (super)
-
stepan authored
Instead of having static variable with this value, it is part of RCodeBuilder and transitively part of RContext.
-
stepan authored
-
stepan authored
The ReplacementBlockNode executes RHS and the target and then executes ReplacementNode, which can specialize on target value. In the default case ReplacementNode creates nodes that handle replacement in the same way as before: SpecialReplacementNode and GenericReplacementNode.
-
stepan authored
Replacements are broken down to individual calls lazily on first execution. This is done by new ReplacementBlockNode, which executes RHS store to tmp variable then the actual replacement and then it returns the RHS value. ReplacementBlockNode creates the node for actual replacement depending on the replacement structure: if all the calls are specials, it creates ReplacementNodeSpecial, which takes advantage of the properties of specials. Otherwise it creates regular replacement node that executes exactly the same function calls as GnuR would.
-
Stepan Sindelar authored
Merge pull request #446 in G/fastr from ~STEPAN.SINDELAR_ORACLE.COM/fastr:feature/fixes-for-jvmtrend2 to master * commit 'b21c05cd': Removed flag forNextMethod from argument matching Tests for NextMethod argument matching Covcor converted to cast pipelines + tests Varargs related methods moved from ArgumentsSignature to new class VarArgsHelper
-
Mick Jordan authored
-
Mick Jordan authored
-
stepan authored
The flag seems to be not needed anymore since we now do the argument matching for NextMethod in more generic way.
-
stepan authored
-
stepan authored
-
stepan authored
-
Mick Jordan authored
-