clean-build issueshttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues2021-04-22T15:45:49+02:00https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/48Gast test scripts still use test0.prj2021-04-22T15:45:49+02:00Camil StapsGast test scripts still use test0.prjThe `test.sh` scripts for Gast still assume test0.prj exists, which has been removed in https://gitlab.science.ru.nl/clean-and-itasks/gast/-/merge_requests/45. We should instead use `make -C Tests` instead, and also something on Windows.The `test.sh` scripts for Gast still assume test0.prj exists, which has been removed in https://gitlab.science.ru.nl/clean-and-itasks/gast/-/merge_requests/45. We should instead use `make -C Tests` instead, and also something on Windows.Erin van der VeenErin van der Veenhttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/47arm64 build is unstable2021-04-23T13:22:51+02:00Mart Lubbersarm64 build is unstableit fails on the language report but if I retrigger manually it builds fineit fails on the language report but if I retrigger manually it builds finehttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/46Build language report separately2021-03-19T15:35:43+01:00Mart LubbersBuild language report separatelyBuild the language report separately so that the nightly build can be linked to from the wikiBuild the language report separately so that the nightly build can be linked to from the wikihttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/45upgrade scripts in Tools/* to python32021-03-19T15:36:08+01:00Mart Lubbersupgrade scripts in Tools/* to python3the buildserver now runs python3the buildserver now runs python3Mart LubbersMart Lubbershttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/38Update Linux versions on build machines2021-07-27T10:02:31+02:00Camil StapsUpdate Linux versions on build machinesThe x64 and x86 Linux VMs for Jenkins are running Ubuntu 16.04, which is EOL coming April. If we have not switched to GitLab for builds by then, we should update to a later version.The x64 and x86 Linux VMs for Jenkins are running Ubuntu 16.04, which is EOL coming April. If we have not switched to GitLab for builds by then, we should update to a later version.Camil StapsCamil Staps2021-05-03https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/37Follow-up from "Add an overlapping instance allowing compiler"2020-10-13T13:49:41+02:00Camil StapsFollow-up from "Add an overlapping instance allowing compiler"The following discussion from !62 should be addressed: https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/merge_requests/62#note_113122
- [x] `sed` could be used instead of `patch` to avoid build failure
- [x] Later, or alterna...The following discussion from !62 should be addressed: https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/merge_requests/62#note_113122
- [x] `sed` could be used instead of `patch` to avoid build failure
- [x] Later, or alternatively, a compiler flag could be used to enable allowance of overlapping instances
- [x] There is an unused `$2` in the linux-x86 build scriptMart LubbersMart Lubbershttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/36Windows build for clean-base broken2019-11-25T09:24:53+01:00Camil StapsWindows build for clean-base brokenThe build for clean-base on x64 windows is broken. Looking at the end of the log:
```
C:\Jenkins\workspace\clean-base-windows-x64\build\objectio>
C:\Jenkins\workspace\clean-base-windows-x64>c:\msys64\usr\bin\bash.exe generic/test.sh cle...The build for clean-base on x64 windows is broken. Looking at the end of the log:
```
C:\Jenkins\workspace\clean-base-windows-x64\build\objectio>
C:\Jenkins\workspace\clean-base-windows-x64>c:\msys64\usr\bin\bash.exe generic/test.sh clean-base windows x64
```
it seems that the script finishes after building ObjectIO, while it should continue to build cpm and rebuild the Clean system ([source](https://gitlab.science.ru.nl/clean-and-itasks/clean-build/blob/72e66a794fe5f714d6ea979f611676d2ee711b44/clean-base/windows-x64/build.sh#L454)).https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/35Heap full when compiling iTask.Engine on windows 64 bit2019-10-21T11:01:10+02:00Rinus PlasmeijerHeap full when compiling iTask.Engine on windows 64 bitThe default heap settings of the clean compiler, 125 MB, was not enough to compile this module out of the box.
200 MB is fine.The default heap settings of the clean compiler, 125 MB, was not enough to compile this module out of the box.
200 MB is fine.Bas LijnseBas Lijnsehttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/34Build failures not recognized on Windows2019-11-25T09:23:59+01:00Camil StapsBuild failures not recognized on WindowsFor instance, http://localhost:8081/job/clean-ide-windows-x64/364/console failed to build the IDE:
```
Error [PmDriver.icl,1805,<node definition>]: gst_world selector not defined
Error [PmDriver.icl,1806,<node definition>]: gst_world no...For instance, http://localhost:8081/job/clean-ide-windows-x64/364/console failed to build the IDE:
```
Error [PmDriver.icl,1805,<node definition>]: gst_world selector not defined
Error [PmDriver.icl,1806,<node definition>]: gst_world not defined as a record field
Error [PmDriver.icl,1809,<node definition>]: gst_world selector not defined
Error [PmDriver.icl,1810,<node definition>]: gst_world not defined as a record field
```
However the job is not marked as failed and so the nightly build does not contain an IDE.https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/33Move to gcc for linking2019-10-21T09:12:32+02:00Camil StapsMove to gcc for linkingIf I'm not mistaken, the Clean optimizing linker does not support exporting a symbol table on Linux, which complicates the interworking features of https://gitlab.science.ru.nl/clean-and-itasks/abc-interpreter.
In my tests, `/usr/bin/gc...If I'm not mistaken, the Clean optimizing linker does not support exporting a symbol table on Linux, which complicates the interworking features of https://gitlab.science.ru.nl/clean-and-itasks/abc-interpreter.
In my tests, `/usr/bin/gcc:--gc-sections` as `EnvironmentLinker` leads to executables of roughly the same size as `lib/exe/linker`. @johnvg, is there any reason why the optimizing linker should still be used?https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/29windows 64 bundle complete is missing2019-07-22T09:52:54+02:00Mart Lubberswindows 64 bundle complete is missingBas LijnseBas Lijnsehttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/28TCP not included in the windows x86 bundle complete2019-10-11T14:25:11+02:00Mart LubbersTCP not included in the windows x86 bundle completeBas LijnseBas Lijnsehttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/27Building for date fails with dependencies2018-09-23T10:47:55+02:00Camil StapsBuilding for date fails with dependenciesIt seems `build_dependency` in `setup.sh` still expects the old directory structure:
```console
camil@hosea:/ho/ca/Ve/clean-build$ CLEANDATE=2018-09-01 ./generic/fetch.sh clean-bundle-complete linux x64
camil@hosea:/ho/ca/Ve/clean-build...It seems `build_dependency` in `setup.sh` still expects the old directory structure:
```console
camil@hosea:/ho/ca/Ve/clean-build$ CLEANDATE=2018-09-01 ./generic/fetch.sh clean-bundle-complete linux x64
camil@hosea:/ho/ca/Ve/clean-build$ CLEANDATE=2018-09-01 ./generic/setup.sh clean-bundle-complete linux x64
Building dependency: clean-base
./generic/setup.sh: line 33: cd: ../../clean-base/linux-x64: No such file or directory
```https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/26Save time and bandwidth on SVN checkouts2018-08-30T11:39:18+02:00Camil StapsSave time and bandwidth on SVN checkoutsEspecially on bundles using `clean-libraries` it takes quite some time to download the repository, while with SVN it is not needed to download everything. There are two ways to speed this up:
1. For instance, checkout `https://svn.cs.ru...Especially on bundles using `clean-libraries` it takes quite some time to download the repository, while with SVN it is not needed to download everything. There are two ways to speed this up:
1. For instance, checkout `https://svn.cs.ru.nl/repos/clean-libraries/trunk/Libraries/TCPIP` instead of `https://svn.cs.ru.nl/repos/clean-libraries/trunk`
2. Use `svn export` instead of `checkout` to discard versioning information
When building `bundle-complete` for a particular date, `clean-libraries` is downloaded many times, and with these changes we can win quite some time there.https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/25clean-base on 64-bit linux requires unsupported relocation 2 for cpm2018-09-29T09:12:20+02:00Camil Stapsclean-base on 64-bit linux requires unsupported relocation 2 for cpmWhen building clean-base on my 64-bit linux install *with* a working Clean installation (#23) it fails to build cpm:
```
Linking Cpm
relocation 2 in data section not supported
```When building clean-base on my 64-bit linux install *with* a working Clean installation (#23) it fails to build cpm:
```
Linking Cpm
relocation 2 in data section not supported
```https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/24clean-base fails with elf_relocations error in Docker image of Debian Stretch2018-09-24T21:51:26+02:00Camil Stapsclean-base fails with elf_relocations error in Docker image of Debian StretchIt is impossible to build clean-base in a Docker image of Debian Stretch with gcc 6.3.0. The linker fails on `cocl` because of a missing case:
```
Linking cocl
Run time error, rule '<case>[line:181];29;63' in module 'elf_relocations' do...It is impossible to build clean-base in a Docker image of Debian Stretch with gcc 6.3.0. The linker fails on `cocl` because of a missing case:
```
Linking cocl
Run time error, rule '<case>[line:181];29;63' in module 'elf_relocations' does not match
```
This is odd, because on my 'normal' installation of Debian Stretch with the same gcc this does not occur.
@johnvg, could you maybe have a look?https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/23clean-base depends on an existing Clean installation2018-09-29T09:12:20+02:00Camil Stapsclean-base depends on an existing Clean installationIt seems impossible to build clean-base on a 64-bit linux machine without Clean installed. After running fetch and setup, building fails at this step:
```
second compile of system modules
Compiling backendconvert
clm: can't start the cl...It seems impossible to build clean-base on a 64-bit linux machine without Clean installed. After running fetch and setup, building fails at this step:
```
second compile of system modules
Compiling backendconvert
clm: can't start the clean compiler: No such file or directory
```
This is in the `build_compiler` function.
Is it intended that this works on systems without Clean installed?https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/22Project files broken for Object IO examples2021-03-25T19:46:13+01:00Mart LubbersProject files broken for Object IO examplesThey assume that the project is in a different folder than it actually is.
It assumes that the examples are in:
`$CLEAN_HOME/Examples/ObjectIO Examples` but they are in `$CLEAN_HOME/Examples/ObjectIO/`They assume that the project is in a different folder than it actually is.
It assumes that the examples are in:
`$CLEAN_HOME/Examples/ObjectIO Examples` but they are in `$CLEAN_HOME/Examples/ObjectIO/`https://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/21basicapiexamples.icl is not generated in the builds2018-10-22T10:25:41+02:00Mart Lubbersbasicapiexamples.icl is not generated in the buildshttps://gitlab.science.ru.nl/clean-and-itasks/clean-build/-/issues/20object io environment is not included in the windows bundle complete2019-10-11T14:25:21+02:00Mart Lubbersobject io environment is not included in the windows bundle complete