21 Feb
2015
21 Feb
'15
11:34 p.m.
Le 21/02/2015 23:24, Johan Fabry a écrit :
So the easy solution in ‘production use’ would be to create the
parser once and then keep hanging on to it? I would not say that this
is really a PetitParser issue, but more how the parser is used.
All other Smalltalk parsers are instantiated as many times as the
PetitParser one in this benchmark, and their instanciation time is
around 0 ms.
But, yes, I would think that one of the solution would be a
pre-instantiation of a PetitParser, and a final instantiation to create
an instance.
In a way, one could imagine writing SmaCC parsers in the same way that
PetitParser parsers are written... pre-instantiation being the part
where SmaCC generates and compile its parsers.
In this benchmark, this would cut parsing time by 2.
Thierry