since the script is so simple, could you try logging directly onto the commander resource that was used to run it originally, and run the script directly from there? I'm trying to figure out if it's something in the commander environment that is tripping up emake.
It was all my fault. We run more on more than one resource but I specified the same paths for both
steps, I separated the xml and db paths and everything is better now.
I don't want the huge log files, how do I lower the verbosity?
Yuri, it looks like there are two types of problems with the annotation file. First, you are using the 'clear' command in the middle of your build for some reason. This emits control characters that are normally interpreted by the terminal to clear the screen. This obviously doesn't make much sense in the context of an automated build. Unfortunately, those control characters are not valid in raw form in an XML document. The easiest thing to do is strip those characters out of the annotation before trying to load it. I will log a bug report for this issue, but I don't know when we will be able to address it.
Second, the annotation appears to be truncated. It just stops shortly after the command that prints "Building: Pci9030". This could be caused by a variety of things -- the build was aborted, or the file was written to a network drive that had connectivity problems, etc.
The compilation aborts with the following error message:
checking whether make sets $(MAKE)... ./configure: eval: line 20417: syntax error near unexpected token `<'
./configure: eval: line 20417: `
I guess the problem is that the build scripts deep inside call make and try to use its output, emake annotates the output which look like garbage to the code reading it.
Yuri, that's an interesting discovery! Hopefully in a future release we'll be able to make SparkBuild emake write annotation to a file the same way that ElectricAccelerator emake does.