How to compile a code..blocks project in a remote mode

[eckard_klotz]

Hello Everybody

Currently we create a lot of software-tests by using Code::Blocks together with UnitTest++, where a test is part of a new Code::Blocks project to create a binary. It is very convenient to create the test by using Code::Blocks. But in the end we have to connect every test with an external test- and requirement-manager tool that should be able to define a list of tests which should be repeated and to start them.

In our project we realise every small test-package (a couple of tests which should be used together) as a single build-target of a project. Thus we may save the cb-project file in our test data-base. Once a test should be done the the test-manager should start Code::Blocks to rebuild the associated build-target and start it. How can this be done ?

Best regards,
                  Eckard Klotz.

[MortenMacFly]

Once a test should be done the the test-manager should start Code::Blocks to rebuild the associated build-target and start it. How can this be done ?

Well you can build projects with C::B from the command line, including to close C::B at the end. What you cannot do (atm) is running a certain target of a project from the command line (not only because this may never return). But this you should do within your testmanager anyways, don't you?

[eckard_klotz]

Hello Morten.

Thanks or your quick reply. In the meanwhile I also found the corresponding chapter in the manual 1.12 CodeBlocks at the command line.
Here I found an example:

codeblocks.exe /na /nd --no-splash-screen --built <name>.cbp --target=’Release’

I understand your answer together with the existing target-argument that it is possible to build only one defined target defined in a project. But it will not be started even this is defined as post-build step. But as you mentioned this can be done by the test-manager also. More important is the possibility to have several build-targets in one project where only one can be chosen in the command-line.

So the process would be:

• 1. Use a c::b-project with several build-targets where one target contains only those tests associated with one requirement
• 2. The test-manager has to know which project file contains which target to test one requirement
• 3. The test-manager has to create a os-specific script-file to start C::B to build the desired targets (one target per command-line)
• 4. Once the targets are created, the test-manager has to use additional command-lines to start them

What do you think?

Best Regards,
                       Eckard.

[MortenMacFly]

But it will not be started even this is defined as post-build step.

Pre-and post build steps are run trough. If you starting applications there, this will work. (I didn't think about that.)

What I meant is that there is nothing like a "--run-target=Foo" command line switch. However, running applications in batch in a build process that may not terminate properly including as a post-build step will also not end the C::B batch build. Therefore it maybe risky to do such if you want everything be full automated. For such to work you'll need a "supervisory"/"monitor" process that would kill and shut down everything if i.e. the build is broken, or the running application hangs.

So the process would be:

• 1. Use a c::b-project with several build-targets where one target contains only those tests associated with one requirement
• 2. The test-manager has to know which project file contains which target to test one requirement
• 3. The test-manager has to create a os-specific script-file to start C::B to build the desired targets (one target per command-line)
• 4. Once the targets are created, the test-manager has to use additional command-lines to start them

Point 4 is arguable (due to my explanation above), but basically: Yes.