aboutsummaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAge
* Link against librt if necessaryHEADmasterGravatar Benjamin Barenblat2014-04-14
| | | | | | | | | For a while, we were very confused why ppamltracer would work happily on some Debian systems but produce dynamic linker errors on other, similar ones. It turns out the difference was glibc versions – if you have glibc <2.17, you need to specify '-lrt' for one of the functions ppamltracer uses. This commit adds an Autoconf macro that adds that flag if necessary; it should improve portability in general as well.
* Ignore 'compile' scriptGravatar Benjamin Barenblat2014-04-14
| | | | | | | The previous commit (ca03d49) added 'AM_PROG_CC_C_O' to 'Makefile.am'. This causes Automake to create an extra script, 'compile', a ‘Wrapper for compilers which do not understand '-c -o'’. As an an autogenerated file, this needs to be ignored.
* added line to make autoconf happy on fedora 20Gravatar Matthew Sottile2014-03-21
|
* Version bump to 0.2.1Gravatar Benjamin Barenblat2014-03-10
|
* Relicense under Modified (3-clause) BSD LicenseGravatar Benjamin Barenblat2014-03-10
|
* Version bump to 0.2.0Gravatar Benjamin Barenblat2014-03-04
|
* Tell Windows users to build with '--disable-shared'Gravatar Benjamin Barenblat2014-03-04
| | | | | | | | | | | As detailed in GitHub issue #1, attempting to build ppamltracer as a shared library on Cygwin fails. I believed commit bfc92ba (‘autotools: Don’t build DLLs on Cygwin by default’) fixed this; it does, but it also disables shared library building on non-Cygwin systems. We need to do a release, so this bug is getting deferred until later; for the time being, the documented workaround is to pass '--disable-shared' to 'configure'.
* Python: Correct use of 'super'Gravatar Benjamin Barenblat2014-02-06
| | | | | | I invoked 'super' incorrectly throughout the error-handling code in ppamltracer-python; however, not testing those errors hid this issue until now.
* Python: Make 'report_name_base' optional in calls to 'Tracer'Gravatar Benjamin Barenblat2014-02-06
|
* Racket: Make 'report-name-base' optional in calls to 'call-with-tracer'Gravatar Benjamin Barenblat2014-02-06
|
* Java: Add 'EnvironmentConfiguredTracer'Gravatar Benjamin Barenblat2014-02-06
| | | | | | | | | | | | | | Adding 'ppaml_tracer_init_from_env' (commit b9dfb8d0) makes the Java code substantially more complicated, as it introduces a second mechanism to construct 'Tracer's. While I could simply provide two constructors in the 'Tracer' class, this would make for an ugly and dangerous implementation; after all, the life cycle of a tracer configured explicitly is noticeably different from the life cycle of a tracer configured by sourcing environment variables. Instead, I’ve changed 'Tracer' to be an interface implemented by 'ExplicitlyConfiguredTracer' and 'EnvironmentConfiguredTracer'. This is an API-breaking change.
* Haskell: Add 'withTracer''Gravatar Benjamin Barenblat2014-01-27
| | | | 'withTracer'' wraps 'ppaml_tracer_init_from_env' in the C library.
* Add Description field to pkg-config fileGravatar Benjamin Barenblat2014-01-27
| | | | | Without a Description field, pkg-config complains and won’t provide the library flags for the package.
* Haskell: Fix documentation misspellingGravatar Benjamin Barenblat2014-01-27
|
* Add 'ppaml_tracer_init_from_env'Gravatar Benjamin Barenblat2014-01-27
| | | | | | | | | This new API function is similar to 'ppaml_tracer_init', but instead of accepting the report base name as an argument, it pulls it from the 'PPAMLTRACER_TRACE_BASE' environment variable. This is useful for users who want to be able to vary the locations of trace reports but do not want to build tracing into their configuration file / command-line argument parsers.
* Functional tests for C libraryGravatar Benjamin Barenblat2014-01-27
| | | | | | | | While gearing up to make an addition to the C library, I became concerned at the lack of test infrastructure for the project. The build system now integrates Check <http://check.sourceforge.net/>, a lightweight testing framework for C, and there are a few tests to ensure no serious regressions occur. You can run the tests with “make check”.
* Correct license block in src/timing.cGravatar Benjamin Barenblat2014-01-23
| | | | | | | The license block in src/timing.c suggested that it was not GPL-licensed. That is, however, incorrect; since the release of ppamltracer, src/timing.c has been licensed under the GPLv3+. The license block is now corrected.
* autotools: Don’t build DLLs on Cygwin by defaultGravatar Benjamin Barenblat2014-01-17
| | | | | | | | | | By default, OTF does not build DLLs on Cygwin. Consequently, users who built OTF and ppamltracer using the default 'configure' flags received warnings about missing DLLs during ppamltracer’s link phase. This commit changes that behavior, causing the generated configure script to default to only building static libraries in Cygwin. Should users wish to build DLLs, they can still do so by passing “--enable-shared” to configure.
* Update READMEGravatar Matthew Sottile2014-01-13
|
* Initial commitGravatar Benjamin Barenblat2014-01-13