As long as it tries to process other rules, i.e. first rule has an antlr
error, go onto next rule, I'd be interested to see the speed difference.
However if we are going to explore this sort of thing we need to expose
the different levels in a windows->preferences section, so users can choose.
Could also be worth profiling some rule building, to see where the
bottlenecks are.
Mark
Michael Neale wrote:
Whoops.. hit send...
In drools 3, we have a parser that recovers as best it can, and keeps on
parsing etc... the down side is if there are a large number of rules, with
potentially a large number of errors, performance can be a bit slow while it
tries to recover from all the errors, and fails etc etc...
My question, should have have some upper limit on the occurences of errors,
after which point the rule is just "bad" and marked as such?
Also, if we pick up an error when parsing, should be not try and "compile"
it? (as compiling stuff that isn't parsed fully can yield sometimes annoying
errors).
On 4/8/06, Michael Neale <[EMAIL PROTECTED]> wrote:
