summaryrefslogtreecommitdiff
path: root/src
Commit message (Collapse)AuthorAgeFilesLines
* rebar_xref: ignore opaque type match Dialyzer warningTuncer Ayaz2018-01-301-0/+2
|
* rebar_cover: ignore opaque type match Dialyzer warningTuncer Ayaz2018-01-301-0/+2
|
* rebar_utils: fix Dialyzer warningTuncer Ayaz2018-01-301-1/+1
|
* Log out success message with newlinesJulius Andrikonis2018-01-291-1/+1
| | | | | The raw log file is split into lines to check for errors. However the logged message should contain newlines for easier reading.
* Revert "Drop crypto app reliance"Tuncer Ayaz2017-08-203-10/+10
|
* Merge pull request #642 from tuncer/profile-logLuis Rascão2017-08-191-3/+0
|\ | | | | Remove profiler note
| * Remove profiler noteTuncer Ayaz2017-08-191-3/+0
| | | | | | | | | | If and when getopt gains the required feature, this can be revisited, but let's remove the message, since it's been a long time.
* | Improve pseudo-unique ct_run node nameTuncer Ayaz2017-08-191-3/+9
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Replacing crypto:rand_uniform(0, 10000) with rebar_rnd:uniform(10000) (which is either rand:uniform or random:uniform dependent on what otp version it runs on) fails in rebar_ct. One might assume it's because the old call requests a number >= 0 while the new call has no way to ask for anything but >= 1. However, the way the value is used doesn't rely on it to include 0, so the bug lies elsewhere. Actually, generating a random number to choose a pseudo-unique ct_run node name isn't a reliable method to avoid naming clashes. I cannot reproduce it locally and also cannot see the actual error in the CI logs. Mirroring what rebar_eunit does and using a bigger rng range seems to be enough to avoid node name clashes on Travis-CI.
* | Drop crypto app relianceTuncer Ayaz2017-08-183-8/+2
|/ | | | | | | | | | By doing this, we fix two issues at once. First, there's no need to have crypto available anymore. While not having crypto in your Erlang installation is a questionable packaging decision, it tends to happen in the wild with users installing Erlang and missing crypto. Sometimes this is not due to a distro's package but users building Erlang without the needed OpenSSL dependencies. Second, this resolves the Erlang 20 rng deprecation warnings.
* Merge pull request #636 from tuncer/fix-634Luis Rascão2017-08-181-1/+7
|\ | | | | Fix #634
| * Fix #634Tuncer Ayaz2017-08-181-1/+7
| |
* | Properly skip apps with a .app.src.script filePaul J. Davis2017-05-151-2/+2
|/ | | | | | Any application using a .app.src.script file will now be properly skipped if requested. This check was missing the newer third case for the file extension.
* Fix typo in eunit help.Pete Higgins2017-02-211-2/+2
|
* Remove superfluous space in unabbrev error msgTuncer Ayaz2016-09-161-1/+1
|
* port_compiler: fix handling of multiple specsTuncer Ayaz2016-08-301-6/+6
|
* rebar_utils: restore blank lineTuncer Ayaz2016-08-281-0/+1
|
* Fix cdb processing when a file is skippedTuncer Ayaz2016-08-281-3/+5
|
* Merge pull request #617 from tuncer/linkcppLuis Rascão2016-08-261-13/+33
|\ | | | | port_compiler: auto-select C++ specific link template
| * port_compiler: auto-select C++ specific link templateTuncer Ayaz2016-08-261-13/+33
| | | | | | | | | | | | | | | | Even though "$CC foo.o -lstdc++ -o foo.so" works if foo.o was built with $CXX, depending on the platform and toolchain and C++ standard used, it may not. In order to ensure this works reliably, we introduce EXE_LINK_CXX_TEMPLATE and DRV_LINK_CXX_TEMPLATE and auto-select if the port spec being built consists of at least one C++ source.
* | port_compiler: clean up compile_each()Tuncer Ayaz2016-08-261-21/+25
|/ | | | Extract clang db entry generation into a separate function.
* port_compiler: generate clang compilation dbTuncer Ayaz2016-08-191-13/+51
| | | | | | In order for newer clang tools to work, they require the presence of a compilation database in the form of compile_commands.json. Therefore, we adapt port_compiler to write such a file.
* Introduce and use compat random number moduleTuncer Ayaz2016-08-063-2/+183
| | | | | | | | In order for rebar to work with previous and current OTP releases, we generate the rebar_rnd module on startup. rebar_rnd is generated dynamically to make sure that we use the right random module (either rand if available or else random). It only wraps the common subset of the API, but that's sufficient for rebar's use.
* port_compiler: consistently format default_env/0Tuncer Ayaz2016-07-171-2/+2
|
* qc: lift experimental noticeTuncer Ayaz2016-07-121-4/+1
|
* Merge pull request #606 from lrascao/fix/revert_590Luis Rascão2016-06-081-3/+2
|\ | | | | Revert 'Treat port env vars as expandable only if they self reference'
| * Revert 'Treat port env vars as expandable only if they self reference'Luis Rascão2016-06-071-3/+2
| | | | | | | | Regression introduced in b816c69e343c8fd757c59fe8703eeda597f4da0a.
* | Add necessary missing OSX flags for port driver linkingLuis Rascão2016-06-071-3/+3
|/
* Add raw to VsnRegex-Free depJesse Gumm2016-03-201-0/+5
|
* Support rebar3-style depsJesse Gumm2016-03-201-0/+8
| | | | | | | | Example: {git, {appname, "git://something/something", {branch, master}}} ^ | |---- Notice the lack of a version Regex
* Merge pull request #588 from lrascao/fix/look_for_ct_spec_files_only_in_test_dirLuis Rascão2016-03-181-4/+16
|\ | | | | Look for ct .spec files in the ct_dir that was specified
| * Optionally look for ct .spec files in the ct_dir that was specifiedLuis Rascão2016-03-141-4/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | Allow to change existing behaviour which is to find all .spec files recursively in the current working directory through a new rebar.config option: {ct_search_specs_from_test_dir, true} This is confusing since the user explicitly stated the location for his spec files and negates the possibility to have different spec'ed suites for different environment. Also fix the node name generation on the ct test run, append a random number that will reduce chances of name collisions on epmd.
* | Introduce REBAR_VSN_CACHE_FILE env variable to load/save vsn cacheMatwey V. Kornilov2016-03-183-3/+24
| | | | | | | | | | | | | | | | | | | | | | | | When REBAR_VSN_CACHE_FILE is set, then vsn cache is loaded from this file on rebar start and updated when new data. Under specific circumstances (i.e. in build environments), full git tree may not be available, but only its snapshot. We need a way to use preheat vsn cache instead of invocing git command. Signed-off-by: Matwey V. Kornilov <matwey.kornilov@gmail.com>
* | Introduce REBAR_DEPS_PREFER_LIBS env variable to alter search behaviourMatwey V. Kornilov2016-03-142-15/+42
| | | | | | | | | | | | | | | | | | | | | | When REBAR_DEPS_PREFER_LIBS is set, dependencies with defined sources are allowed to be searched for in system lib directory. Under specific circumstances (i.e. in build environments without networking) it is impossible to fetch deps locally. So, user needs a way to ask rebar to search in system lib directory as well. Signed-off-by: Matwey V. Kornilov <matwey.kornilov@gmail.com>
* | Fix bug when running gcc in cross_sizeofFrank Hunleth2016-03-131-1/+1
|/ | | | | | | The test program used to determine the word size of a crosscompiler is crafted to return an error, so this changes the logic to expect an error exit. If the crosscompiler actually compiles the test program successfully, that would be remarkable and worthy of investigation.
* Merge pull request #590 from lrascao/feature/allow_port_env_flags_redefinitionLuis Rascão2016-02-231-2/+8
|\ | | | | Treat port env vars as expandable only if they self reference
| * Treat port env vars as expandable only if they self referenceLuis Rascão2016-02-191-2/+8
| | | | | | | | | | Provide additional port test case. Also, update newest OS X versions build flags.
* | Fix eunit_surefire crash with rebar eunit tests=...Tomas Janousek2016-02-191-1/+6
|/ | | | | | | This fixes the following crash: > =ERROR REPORT==== 20-Jun-2014::16:35:38 === > Error in process <0.40.0> with exit value: {function_clause,[{eunit_surefire,escape_suitename,[undefined,[]],[{file,"eunit_surefire.erl"},{line,416}]},{eunit_surefire,write_report,2,[{file,"eunit_surefire.erl"},{line,258}]},{lists,foreach,2,[{file,"lists.erl"},{line,1336}]},{...
* Update regexp to account for newer OTP versionsLuis Rascão2016-02-151-7/+13
| | | | | | | | Add support for patch numbers, also expose version_tuple/1 for easier internal access. Normalize version_tuple/1 return value, now always {Major, Minor, Patch}. Update regression tests accordingly.
* Ensure ebin created for dia compilerYury Gargay2016-02-131-0/+2
| | | | also add myself to THANKS
* Automatically clean neotoma-generated erl filesJohn R. Daily2016-02-131-1/+32
| | | | Add myself to THANKS
* Add support for Windows integration testingLuis Rascão2016-01-151-9/+26
| | | | | | | | | | | | | | Use retest feature/rebar-windows-ci branch that adds Windows tests support, test setup callback and additional touch command. For all tests copy rebar and rebar.cmd using retest setup callback. Port OS specific commands used in tests to Erlang (eg. touch, rm, cp, stat..). rebar_ct: do away with grep command line invocation (which doesn't exist in Windows) and use instead plain Erlang parsing. Increase timeout for rgen1 test to 4 minutes, Windows Appveyor can take longer than the previous 2 minutes.
* Merge pull request #573 from lrascao/feature/improve_relup_generation_debugFred Hebert2015-12-131-0/+6
|\ | | | | Provide additional debug logging on relup generation
| * Provide additional debug logging on relup generationLuis Rascão2015-12-131-0/+6
| |
* | Merge pull request #570 from tuncer/xref-dialyzer-warningLuis Rascão2015-11-281-3/+3
|\ \ | |/ |/| xref: fix dialyzer warning introduced in 69802f63120
| * xref: fix dialyzer warning introduced in 69802f63120Tuncer Ayaz2015-11-281-3/+3
| |
* | Merge pull request #557 from lrascao/fix/windows_eunit_testsFred Hebert2015-10-261-5/+7
|\ \ | | | | | | Fix windows eunit tests
| * | Fix windows eunit testsLuis Rascão2015-10-251-5/+7
| |/ | | | | | | | | | | | | | | | | | | | | File tests: windows file operations should abort on error the same as unix operations invoked through sh. Also windows does not support the '?' character in filenames. Eunit tests: the 'All' prefix is missing on the 'x tests passed' message. Eunit only prints the 'All' prefix if there are more than 2 passed tests, dropping the prefix on the match works for all cases.
* | Add qualified name tests docs (see pr #119)Sergey Savenko2015-10-251-0/+1
|/
* fprof: further document Cachegrind supportLuis Rascão2015-10-191-2/+2
|
* fprof: document Cachegrind supportTuncer Ayaz2015-10-141-1/+3
|
View Lorry Controller Status