On Fri, Jun 23, 2006 at 04:01:10PM -0500, Satya wrote: > >Basically, it allows bison to generate a parser that parses 1 token at a > >time. You feed it the tokens one at a time, and it updates the parser > >until you've parsed everything. > > Really good! Such a parsing algorithm can be used by the IDE to > simulate a test-parse; so you interface the parser with a parse > context I see. This tool called 'Anagram' also implements such an > Idea. They call it an 'event driven' parser. In Anagram, you can > communicate with the parser through a data stucture called a PCB > (parser control block);
Yes, that makes sense. I used to use the term asynchronous parser. I needed this feature to parse some data coming over a pipe. When I brought up this type of parser, the experienced bison developers were using the term push parser, which is how I started to use this term. If you search the web, you will see that push/pull are standard terms to describe these 2 types or parsers. The patch is probably 90% complete and 100% working.I have to fix up the interface the user will see, a few timing improvements, code cleanups, ... If you are interested in using it, you might be interested in trying it out to see if it suits your needs before it gets put into the mainline. It will certainly work for what i need. :) Bob Rossi
