Thread: pgsql: Make pgbench's expression lexer reentrant.
Make pgbench's expression lexer reentrant. This is a necessary preliminary step for making it play with psqlscan.l given the way I set up the lexer input-buffer sharing mechanism in commit 0ea9efbe9ec1bf07. I've not tried to make it *actually* reentrant; there's still some static variables laying about. But flex thinks it's reentrant, and that's what counts. In support of that, fix exprparse.y to pass through the yyscan_t from the caller. Also do some minor code beautification, like not casting away const. Branch ------ master Details ------- http://git.postgresql.org/pg/commitdiff/429ee5a822db0e8faf669d77c810f1eeaaff1ab4 Modified Files -------------- src/bin/pgbench/exprparse.y | 57 ++++++++++++++++++++----------- src/bin/pgbench/exprscan.l | 83 ++++++++++++++++++++++++++++----------------- src/bin/pgbench/pgbench.c | 15 +++++--- src/bin/pgbench/pgbench.h | 26 +++++++++----- 4 files changed, 117 insertions(+), 64 deletions(-)
Hi Tom, On 2016-03-19 20:35:48 +0000, Tom Lane wrote: > Make pgbench's expression lexer reentrant. > > This is a necessary preliminary step for making it play with psqlscan.l > given the way I set up the lexer input-buffer sharing mechanism in commit > 0ea9efbe9ec1bf07. > > I've not tried to make it *actually* reentrant; there's still some static > variables laying about. But flex thinks it's reentrant, and that's what > counts. > > In support of that, fix exprparse.y to pass through the yyscan_t from the > caller. Also do some minor code beautification, like not casting away > const. This appears to have caused build failures on gaur, prairiedog, and pademelon http://pgbuildfarm.org/cgi-bin/show_log.pl?nm=gaur&dt=2016-03-20%2005%3A53%3A21 http://pgbuildfarm.org/cgi-bin/show_log.pl?nm=prairiedog&dt=2016-03-19%2023%3A21%3A23 http://pgbuildfarm.org/cgi-bin/show_log.pl?nm=pademelon&dt=2016-03-20%2007%3A59%3A38 Not sure you saw those already. - Andres
Andres Freund <andres@anarazel.de> writes:
> On 2016-03-19 20:35:48 +0000, Tom Lane wrote:
>> Make pgbench's expression lexer reentrant.
> This appears to have caused build failures on gaur, prairiedog, and
> pademelon
Yeah, I know. I have a plan for fixing it but wanted to get the
main pgbench parsing rewrite pushed first.
The problem seems to be that the ancient bison version these critters
are using has a bug for the combination of a reentrant lexer and a
non-reentrant parser. Which it's not so surprising they'd not have
tested. I think it won't be too painful to dodge this by making the
parser reentrant too (bison %pure-parser), and will look into that
this evening.
regards, tom lane