On Mon, Aug 19, 2024 at 12:47:02PM +0200, Pablo Neira Ayuso wrote: > Hi Phil, > > On Fri, Aug 16, 2024 at 02:25:12PM +0200, Phil Sutter wrote: > > On Wed, Aug 14, 2024 at 09:26:36PM +0200, Phil Sutter wrote: > > [...] > > > Maybe one could introduce a start condition which allows strings, but > > > it might turn into a mess given the wide use of them. I'll give it a try > > > and let you know. > > > > Looks like I hit a dead end there: For expressions like 'iif', we have > > to accept STRING on RHS and since I need a token to push SC_STRING, I > > can't just enable it for all relational expressions. The alternative is > > to enable it for the whole rule but I can't disable it selectively (as I > > had to enable it again afterwards without knowing what's next. :( > > flex rules also tells what to find first (order implies priority) > maybe a combination of start conditions to carefully placing. I can > take my poor man fix by now so this can be revisited later :) I have a working draft using an exclusive start condition (Florian pointed me to that). It passes the testsuite, might need more work though (the first token after the limit statement is still parsed in the exclusive condition, so all statement openers must potentially be valid in all conditions. Anyway, it's not a simple fix from my side so please go ahead and we'll discuss my patches later as the "academic project" they are. Cheers, Phil
