- 21 12月, 2016 6 次提交
-
-
由 Matt Ellis 提交于
For cases where git isn't on the path, git is an older version that doesn't support all the flags we need or the repository we are building is not actually a git repository.
-
由 Matt Ellis 提交于
- Sending telemetry isn't really interesting for the CI build. - Restoring a local package cache also isn't useful.
-
由 Matt Ellis 提交于
In the general case, finding the correct crossgen to invoke is non trivial since there's no strong corespondence between the host Runtime ID (which is easy to compute) and the RID that crossgen would be packaged in. For example both RHEL and Centos have crossgen packaged in a nuget packages that uses `rhel.7-x64` as the RID, in order to share the package across minor versions of RHEL and Centos (which is binary compatibile). Since we can build Roslyn without crossgening the bootstrapped toolchain, I'm exposing an option to allow skipping of crossgen for platforms where a developer knows apriori that `crossgen.sh` would not compute the correct RID and fail.
-
由 Matt Ellis 提交于
-
由 Matt Ellis 提交于
-
由 Matt Ellis 提交于
-
- 28 7月, 2016 1 次提交
-
-
由 Matt Ellis 提交于
The --release flag was being ignored to cibuild.sh because the define we pass to Make was set before we actually checked the command line arguments to cibuild.sh. There was also a casing typo in a variable name the script was using which caused the BUILD_CONFIGURATION setting to never be passed when building the non bootstrap projects.
-
- 18 3月, 2016 1 次提交
-
-
由 Jared Parsons 提交于
This was used to deal with unpredictable crashes of MSBuild. Now that we're running under CoreCLR want to surface these errors and give feedback to the appropriate team.
-
- 26 2月, 2016 1 次提交
-
-
由 Jason Malinowski 提交于
By passing --pretty=raw we get the full SHA1s of all parent commits, as well as the tree SHA1 so we can confirm content as well.
-
- 24 2月, 2016 1 次提交
-
-
由 Peter Jas 提交于
In #8701, a comparison was added for `bsd`. It worked on NetBSD as bash returns `netbsd` for `echo $OSTYPE`. However, today I discovered that on FreeBSD, it returns `FreeBSD` with upper cased `BSD`. This change adds the mixed casing to avoid such problems in future for other BSD-like OSes.
-
- 12 2月, 2016 1 次提交
-
-
由 Anders Jensen-Waud 提交于
Add FreeBSD support to build file so that it picks the right path to the GNU make binary (GNU make, which is called ``gmake'' on FreeBSD in order to avoid confusion with BSD Make). The commit also fixes script references to bash with ``env'' in order to make it cross platform (e.g. bash sits in /usr/local/bin/bash not /bin/bash on FreeBSD).
-
- 25 12月, 2015 1 次提交
-
-
由 Jared Parsons 提交于
-
- 24 12月, 2015 3 次提交
-
-
由 Jared Parsons 提交于
-
由 Jared Parsons 提交于
-
由 Jared Parsons 提交于
The makefile wasn’t flowing the custom mono path through to the tools we run during the build.
-
- 22 12月, 2015 1 次提交
-
-
由 Tomas Matousek 提交于
-
- 16 12月, 2015 1 次提交
-
-
由 Tomas Matousek 提交于
-
- 14 12月, 2015 1 次提交
-
-
由 Andy Gocke 提交于
-
- 12 12月, 2015 1 次提交
-
-
由 Andy Gocke 提交于
-
- 08 12月, 2015 1 次提交
-
-
由 Jonathon Marolf 提交于
-
- 03 12月, 2015 1 次提交
-
-
由 Jared Parsons 提交于
The LTTNG library version used in CoreCLR requires LTTNG_HOME be set else it issues a warning on process startup. This was cluttering up our linux build output. Fixing this by exporting the value.
-
- 24 11月, 2015 1 次提交
-
-
由 Jared Parsons 提交于
The first phase of Linux / Mac builds is to build the csc / vbc executables used to build the repo. That initial sttep must use csc / vbc to build our compilers. Previously this was just using our NuGet distributed compilers under Mono. Now it uses our compilers running on coreclr. Ideally this would be the same compilers as our NuGet packages. That's not quite available today. Hence this is built off a spefici SHA instead.
-
- 22 11月, 2015 1 次提交
-
-
由 Jared Parsons 提交于
This changes the Mac builds to use coreclr for running csc and vbc
-
- 19 11月, 2015 1 次提交
-
-
由 David Kean 提交于
-
- 18 11月, 2015 4 次提交
-
-
由 Andy Gocke 提交于
-
由 Andy Gocke 提交于
-
由 David Kean 提交于
-
由 Jared Parsons 提交于
This changes the compilers and server to exit the process when a Debug.Assert call fails in a bootstrap build. The current behavior of throwing up a dialog blocks the Jenkins build until a timeout occurs and leaves no data. Now a full stack trace will be printed, the process exits with a non-zero exit code and the entire run will fail promptly.
-
- 17 11月, 2015 1 次提交
-
-
由 David Kean 提交于
This includes the CPS SDK so that C#/VB CPS implementation can be merged into future.
-
- 08 11月, 2015 1 次提交
-
-
由 Jared Parsons 提交于
-
- 06 11月, 2015 1 次提交
-
-
由 David Kean 提交于
-
- 04 11月, 2015 2 次提交
-
-
由 Tomas Matousek 提交于
-
由 Neal Gafter 提交于
-
- 03 11月, 2015 2 次提交
-
-
由 Jared Parsons 提交于
-
由 Tomas Matousek 提交于
-
- 31 10月, 2015 1 次提交
-
-
由 Andy Gocke 提交于
-
- 29 10月, 2015 1 次提交
-
-
由 Andy Gocke 提交于
-
- 28 10月, 2015 1 次提交
-
-
由 Tomas Matousek 提交于
-
- 23 10月, 2015 1 次提交
-
-
由 Andy Gocke 提交于
-
- 22 10月, 2015 1 次提交
-
-
由 Andy Gocke 提交于
-