Parsing spt test in unstable extremely slow and CPU intensive
Syntax highlighting a single SPT in eclipse takes minutes (even on small files) and CPU load of java increases upto 367%
Running single tests takes even longer and can be considered unusuable.Introduced since upgrade to the unstable spoofax version (as recommended in the Compiler Construction lab)
Version according to eclipse:
Submitted on 26 September 2013 at 17:41
Spoofax/IMP 1.2.0.0-s40497 org.strategoxt.imp.feature.group null
Issue Log
I have a few questions:
- Does this happen in a workspace, where you only have your test project?
- Is parsing in the generated MiniJava editor fast?
- Do you get warnings about ambiguous fragments in the MiniJava editor?
- I have only one project, which contains both my language definition and my test cases (as was recommended by one of the lab assistants).
- I will test this asap and report on it.
- At this moment no such warnings are reported; but I would be suprised if my language definition is complete unambiguous as I have not yet defined operator precedence for example.
That was actually a bad advice, because the lab description tells you to separate them. W.r.t. ambiguity, I was referring more to ambiguous layout. However, I think we had an idea what is causing this.
I moved the tests back to it’s own project and made sure it was the only project open in Eclipse, but I observe the same behaviour.
Right now I can’t test the Minijava Editor responsiveness/warnings about ambiguous layout.
If you still need me to test that, I’ll do so ASAP.
The current issue is the amount of test cases. You can reduce this, by simply closing the test project. This allows you to work on your grammar, but then you cannot test it. If you want to test it, you might spread your test cases over several projects for now and open them as needed. We know this is not the perfect situation, but we are investigating the root cause for this.
The ambiguity was just a typical reason why such things happen. But here it seems to be on our site.
Oke, thank you for the workaround.
Do you need my project or have you been able to replicate the behaviour?
Additionally: This did not occur for all students yesterday. I know of atleast 1 person who had a similar test setup, with a similar amount of tests that did not have this problem; on a slower machine.
Log in to post comments