In r23492, reference resolution tests fail with a stack trace mentioning editor-analyze, when your analysis strategy is called editor-analyse (with an s instead of a z). Reason might be an hard encoded editor-analyze, instead of using the observer mentioned in the editor description?

Submitted by Guido Wachsmuth on 1 November 2011 at 17:53

On 2 November 2011 at 11:38 Lennart Kats commented:

AFAIK it gets the observer through spt-get-observer in https://svn.strategoxt.org/repos/StrategoXT/spoofax-imp/trunk/org.strategoxt.imp.testing/trans/spoofax_testing.str. Maybe you specified multiple observers in the editor descriptors, including one called editor-analyze?


On 3 November 2011 at 17:24 Guido Wachsmuth commented:

The strategy seems to be the editor-analyze from the testing language, not from the Spoofax language. A student pointed out, that this happens when your reference resolution test has errors and/or warnings (wrt the tested language). Resolution tests without errors/warnings seem to work.


On 3 November 2011 at 17:24 Guido Wachsmuth tagged lab

On 3 November 2011 at 17:24 Guido Wachsmuth tagged testing

On 3 November 2011 at 17:45 Lennart Kats commented:

Oh. Am I understanding correctly that you just get a stack trace when the tested language’s reference resolver fails? Or is it reporting a stack trace when there are semantic errors/warnings in the test input? Either way it sounds like a presentational issue?


On 3 November 2011 at 18:59 Guido Wachsmuth commented:

The test case should be correct (w.r.t. reference resolution), but is marked with an error (or warning?). The error text includes the stack trace, saying Unexpected error: rewriting failed editor_analyze_0_0 editor_analyze_0_0 scope_paths_1_2 dr_scope_1_1 dr_scope_1_1 collect_all_2_0 foldr_3_0 foldr_3_0 check_warning_0_0 filter_1_0 check_expectation_0_3 map_1_0 execute_service_0_1 editor_resolve_0_0 resolve_0_0


On 4 November 2011 at 11:21 Lennart Kats commented:

Ok, but that should really be a warning. The only issue is then that it’s reporting a small part of the testing language’s stack trace as well as the trace of the language under test. I changed the message to “unexpected error resolving” now, if that helps.


On 4 November 2011 at 11:21 Lennart Kats tagged minor

On 4 November 2011 at 11:44 Vlad Vergu commented:

I’ve seen a case like this on Tuesday (not sure if the same as Guido’s). The test hover displayed a stack trace that did not seem to include any target language strategies (i.e. only testing language). The reference resolution was working correctly if the fragment was copy-pasted into an empty file and manually exercised.


On 14 November 2014 at 15:07 Daco Harkes commented:

In the current Spoofax version reference resolution tests work fine. Closing issue.


On 14 November 2014 at 15:07 Daco Harkes closed this issue.

Log in to post comments