On Oct 24, 2007, at 9:52 PM, Allan Odgaard wrote:
The WIP C/C++ grammar is matching the full structure of your source code, there are 3 “concerns”:
- It is more fragile (since an unforeseen or wrongly matched
structure can throw it off). 2. It is somewhat slower than the regular grammar. 3. It doesn’t have as good support for the #if 0/1 … #else … #endif as the original grammar (since that structure is orthogonal with the structure of the actual C/C++ code).
Number 1 just means it needs testing to the point where I am confident it works :)
As for #2 and #3, for #2 we can only wait for TM 2.0 to really get that addressed and #3 either requires a lot of duplication in the grammar or also that we wait for TM 2.0 -- but as there is no ETA on TM 2.0, the question is if the demand for the improvements outweigh the tradeoffs, we haven’t really decided that, in the end I guess it is my call, and I am thinking we should make it trunk maybe in 1-2 months.
As a comment on the WIP C syntax: it seems to mark declarations as well as definitions as functions. Not as big of a deal as missing definitions, but probably not ultimately desirable. E.g.:
int testfn(int a); int testfn(int a) { ... }
leads to two "testfn" listings in the symbol menu. This may occur when one has static functions, etc.
Thanks, Matt