Approved. This is better than the status quo, but I think this really boils down to the age-old `eval-when` issue. Without something like eval-when, the compiler can't really figure out the applicability of the switch tag.
Another approach would be to use a whitelist instead of a blacklist -- to only look for switch at the top level or in class definitions (are there any other places it could be meaningfully used)? On 2010-04-12, at 14:43, Henry Minsky wrote: > Change 20100412-hqm-e by [email protected] on 2010-04-12 14:22:14 EDT > in /Users/hqm/openlaszlo/trunk1 > for http://svn.openlaszlo.org/openlaszlo/trunk > > Summary: don't evaluate <switch> inside of literal dataset content > > New Features: > > Bugs Fixed: LPP-3412 > > Technical Reviewer: ptw > QA Reviewer: andre > Doc Reviewer: (pending) > > Documentation: > > Release Notes: > > Overview: > > > Details: > > Parser will ignore <switch> and <include> tags which occur inside of > <dataset> element. > > Note, this check only looks for an ancestor tag named "dataset", so it > won't work in cases where a subclass of <dataset> is being used. > > Tests: > > test/lfc/data/alldata.lzx > smokecheck > amazon > > Files: > M WEB-INF/lps/server/src/org/openlaszlo/compiler/Parser.java > > > Changeset: http://svn.openlaszlo.org/openlaszlo/patches/20100412-hqm-e.tar >
