I have an ant project with over 100 modules. I cycle through all modules compile, package, and publish in one build run. However, when one ivy:publish fails (due to random connection issue), the entire build exits.
I would like the build process to continue compile/publish the remaining modules even if one module fails to publish for whatever reason.
Is there some settings in ivy:publish to prevent exiting upon error or some other way to achieve this?
thanks
Since you appear to be using ANT to call multiple sub-builds, then I would submit this is a control loop problem rather that something specific to ivy. In other words you are best advised to ensure each module's build is as stand-alone as you can make them and then in your loop each module's build should succeed or fail.
You have not indicated what your main build file looks like? I would high recommend using the subant task, as this has a "failonerror" flag that will give you your desired behaviour (build will continue on if a module fails).
<subant failonerror="true">
<fileset dir="." includes="**/build.xml" excludes="build.xml"/>
<target name="clean"/>
<target name="build"/>
</subant>
This should be enough to solve your problem. Any build that fails can be manually re-run. In practice this might be difficult since one module failing might cause a subsequent build to fail due to missing dependencies..... You need to judge the risks of this for yourself.
You can even further complicate your solution later, by using an embedded script to run module builds. If you have lots and lots of errors you might want to add some bespoke error handling logic.
Move a ant dir project after the ant or subant task completes
Related
I am experimenting with the framework to build an SPA using polymer. This will include a large number of custom elements at various levels in the overall application hierarchy. I would like to use web-component-tester to run the module tests on them.
web-component-tester seems to be opinionated about where the tests will be stored - in a separate test directory, where it will run all files found.
I am of an opposite opinion. I would like to store tests in the same directory as the element definition. I would like to differentiate tests by naming them xxx.test.html (or possibly xxx.test.js). I also want to run different "sets" of tests controlled by gulp some of which will be watching for changes and then running the tests (for the app side of my project) and some of which will be elements that use core-ajax to unit test my server side scripts. These will more than likely be in a totally different directory hierarchy (my dist directory) and will be served by a proper web server.
I "think" the "suite" config option wct-conf.js file in my project root might be how I can define this, or alternatively a wct command with some file globs. Unfortunately web-component-tester's README is somewhat confusing on any detail and when you have your own web server it says "You'll need to save WCT's browser.js in order to go this route." What does that mean?
Can someone enlighten me on how can get WCT to run each of the elements/**/*.test.html files as its own "suite" ( I actually intend to use describe, it format - but I assume I still need to use the term suite).
Can someone also explain what I need to do the browser.js when I have my own web server.
I ran some experiments and did a bit of debugging with node-inspector. Firstly, the command line overwrites the suites parameter in the config file
wct app/elements**/*.test.html
does find all my module tests if I have them stored with the elements and ignores the contents of the wct.conf.js file's suites parameter.
also putting the same value (ie app/elements/**/*.test.html) in the wct-conf.js file for the suite parameter does the same job. In fact in this mode, gulp test:local
Also works correctly
So to run different tests for module and distribution, I just need to set up for wct.conf.js for my module tests, and set up gulp to run a command line with the correct location of my test file
I still haven't understood the instructions for running with your own web server.
C#/Mono enables programming/design-by-contract: one can specify "require" and "ensure" statements. This is done with the System.Diagnostics.Contracts library.
If one reads the specifications, it is required to "rewrite" the program before executing it with ccrewrite. This page http://social.msdn.microsoft.com/Forums/en-US/1d0bc7e9-a1bc-469c-982a-47232308f131/ccrewrite seems to suggest that the action must be executed before building the assembly which is strange because the program will be overwritten after building.
What should one do to enable design-by-contract in MonoDevelop?
We have noticed an issue recently that redeployed SSIS packages sometime don't seem to include the latest changes... When I search the dtsx using notepad I see the amended script in the code so the changes are definitely there.
My assumption was that script components of SSIS packages are eventually compiled into an assembly somewhere in the process - this is quite likely since I would imagine C# code cannot run without something compiling it first. So in theory if these assemblies would then end up being cached and not immediately overwritten (for some reason) that would explain this issue.
The only "evidence" that makes me think that my theory is correct is if I keep running the package at some point it suddenly shifts to the new code.
However, so far I haven't found why and how this is happening, if is... Can anybody help?
UPDATE:
MSDN says: "Unlike earlier versions where you could indicate whether the scripts were precompiled, all scripts are precompiled in SQL Server 2008 Integration Services (SSIS) and later versions." - If by pre-compiled they mean that instead of the actual package a pre-compiled version runs (I think this because the package itself does not seem to be compiled since the code is visible in Notepad) there must be a way to force the engine to overwrite the pre-compiled assembly... but how?
UPDATE:
One of the four core components of SSIS is the SQL ServerIntegration Services service, which is a windows service. Apparently this service will cache component/task metadata so that the SSIS runtime engine can poll the cache to see what is installed, which may help speed up package load times. However, if the packages are stored in the file system (not in SQL Integration Services) and executed by Agent Jobs, the agent job will use the 64 bit version of DTEXEC to execute the packages. I haven't yet found evidence that any caching would be involved there, but there are certainly options to check a number of parameters in the validation phase of the execution, such as version numbers - may be for a reason.
Have you looked at sysssispackages to compare the version build number of the package in msdb to your build number in Visual Studio / SSIS?
SELECT name, verbuild
FROM msdb.dbo.sysssispackages
WHERE name LIKE '%bla%'
(Adjust WHERE-clause as necessary to find your package. Do NOT ever "SELECT * FROM msdb.dbo.sysssispackages" as it contains the package XML in one of the columns.)
And in Visual Studio, open the package, then right-click at the background of the package and select "Properties" from the context menu. Look at the field VersionBuild. It should match the number from the SELECT above!
I know this is not an actual solution to your problem but it may help locate where the cause of the problem is. If the number is older, it means that your package deployment did not work.
This sounds somewhat familiar to something I ran into a while back. Unfortunately, I don't remember exactly when I ran into this (so I can't check for sure), but I believe the fix I found was to make sure that I explicitly invoked the Build | Build st_5bd541c294054c25b9e7eb55b92bd0e2 command from the script editor (VSTA) menu before closing the window. (The specific project name will be different for each script, obviously, since it's based on a GUID; however, there will only be one possible submenu under Build.)
Explicitly invoking the Build command ensures that the binary code for the script gets ASCII-encoded and saved in the XML of the resulting .dtsx file. I'd gotten used to SSIS 2005 always building for me whenever I closed the script editor. Apparently, there are bizarre edge cases where SSIS 2008 doesn't always build the script project when the editor closes.
BTW, the precompiled binaries appear to be stored in a tag of the source XML called BinaryItem:
<DTS:Executable DTS:ExecutableType="Microsoft.SqlServer.Dts.Tasks.ScriptTask.ScriptTask, Microsoft.SqlServer.ScriptTask, Version=10.0.0.0, Culture=neutral, PublicKeyToken=89845dcd8080cc91" DTS:ThreadHint="0">
<DTS:Property DTS:Name="ObjectName">SCR_StepOne</DTS:Property>
<DTS:ObjectData>
<ScriptProject Name="ST_5bd541c294054c25b9e7eb55b92bd0e2" VSTAMajorVersion="2" VSTAMinorVersion="1" Language="CSharp" EntryPoint="Main" ReadOnlyVariables="User::FileOneName,User::OutputFolder" ReadWriteVariables="">
<BinaryItem Name="\bin\release\st_5bd541c294054c25b9e7eb55b92bd0e2.csproj.dll">
TVqQAAMAAAAEAAAA//8AALgAAAAAAAAAQAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAgAAAAA4fug4AtAnNIbgBTM0hVGhpcyBwcm9ncmFtIGNhbm5vdCBiZSBydW4gaW4gRE9TIG1v
ZGUuDQ0KJAAAAAAAAABQRQAATAEDADuOb04AAAAAAAAAAOAAAiELAQgAABAAAAAIAAAAAAAAPi8A
AAAgAAAAQAAAAABAAAAgAAAAAgAABAAAAAAAAAAEAAAAAAAAAACAAAAAAgAAAAAAAAMAQIUAABAA
It might be worth checking your source code control system history to see if that was getting updated for some of those screwy errors.
Caveat: I haven't found official Microsoft documentation on this.
This doesn't specifically solve the mystery you have, but if you are running file system-based packages and want to verify that the package that is running is the package you deployed, there is a way to do that.
Build your package.
Open the properties on your package and note down the "Version Build" property (alternatively, open the .dtsx in notepad and find the DTS:VersionBuild attribute.)
Deploy your package.
In your SQL Agent job step, go to the Verification tab.
Enter the Version Build in the "Verify package build" input box.
Execute the job step.
I don't know if this will force SSIS to throw out its cache and get the newly deployed package, but I do know if you modify the .dtsx package's build number by hand and then try to re-run the job step it fails because the package build doesn't match what it's looking for so it is definitely doing a run-time check of that value.
Is there a way to configure Jenkins to resume a (broken) Maven build from where it failed the previous time it ran?
When working with large multi-module projects, it can be very annoying to have to wait a long time for all the unchanged projects to be build after submitting a small change to fix a broken build (e.g. add a missing import after a merge error).
It can of course be done by manually changing the build config to include the -rf <module> in the build-step, but I see at least two problems with this:
I don't want manual edits, it should be automatically.
There might be additional build-steps which I want to skip. (E.g building custom library jars)
I don't think there is a way to tell jenkins to do this automatically, unless there's a plugin that i don't know of that does that.
What you can do to make the process a little more smooth is to create a parameterized build which gets as a parameter the correct module you want to run, so when you click build you could have a drop down menu or something similar to choose from.
I am new at this and I was wondering how I can setup that I save the artifacts, only if less than 90% of the tests have passed.
Any idea how I can do this?
thanks
This is not currently possible with Hudson. What is the motivation to avoid archiving artifacts on every build?
How about a rather simple workaround. You create a post build step (or additional build step) that calls your tests from the command line. Be sure to capture all errors so Hudson don't count it as a failure. Than you evaluate your condition and set the error level accordingly. In addition you need to save reports (probably outside hudson) before you set the error level, so they are available even or only when the build fails.
My assumption here is, that it is OK, not to run the tests when building the app fails. However, you can separate the building and testing in two jobs. See here.