|
From: | josephus |
Subject: | Re: a probalem with Bison |
Date: | Tue, 26 Jul 2005 14:44:54 -0500 |
User-agent: | Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.7.8) Gecko/20050511 |
Paul Eggert wrote:
josephus <address@hidden> writes:I can now parse 3 tokens. the ID token is an error. The grammar says it is not.... the exprssion in the grammar is OWN INTEGER ID this fails when ID is presentd. So I need to decode the tables.Sorry, I can't debug this without more-detailed information. You have sent me some details before, but I couldn't quite follow them and/or reproduce the problem. Perhaps you can try again. It's important to have a complete, self-contained test case that illustrates the problem.I can see where we write the data files. But I need a specification to explain how the tables are used.Unfortunately, there is no external specification for how Bison behaves internally. The only specification is the Bison source code itself. So if you want to proceed in this way, you'll have to look at the source code.I have been working on the debug verison of bison.If you haven't already upgraded, please use: http://www.cs.ucla.edu/~eggert/bison/bison-2.0b.tar.gz I've asked the GNU FTP upload gurus to copy it to here, but it's not there yet: ftp://alpha.gnu.org/gnu/bison/bison-2.0b.tar.gz Here is the MD5 checksum: 47e84c977910be6e136ee6b4bf4967c7 bison-2.0b.tar.gz
I am currently downloading the newest version 20b. I will try it with my problem. It is a complex parse, the input is bison, and the table output is from the input parse. Since you generate the parser from a grammar some of that information is simply not available. I have to reproduce the logic used to build it, to determine if the generation is correct. I need to unravel the parse tables and determine where it went wrong. then I can look at the generation and try to fix it.
I will be leaving for about 3 weeks, when I come back I can work more on the bison source. The version that I sent you will not parse the test.alg. the Makefile does everything. I discovered that bison now automatically copies in the Algol.tab.h. I got double warnings because the older version of bison did not do this. The version i sent will do this. It is a noisy warning but it does not affect the running program.
The grammars I sent you will reproduce the problem I described. The grammar would parse to various different extents, for reasons I cannot identify I reduced the grammar slightly and it quit parsing at all.
I really don't want to work on bison, I want to work on building my compiler.
Flex works just fine and I can build a scan that parses the input file. It was bison that failed. josephus -- Everybody is Ignorant, just on different subjects. --- Will Rogers Jr. It's not what you don't know, it what you know that ain't so. ---Josh Billings
[Prev in Thread] | Current Thread | [Next in Thread] |