[
https://issues.apache.org/jira/browse/THRIFT-775?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12866159#action_12866159
]
Ted Zlatanov commented on THRIFT-775:
-------------------------------------
I didn't attach the HTML report in case you wanted to look at the CSV version
or anything else. I can send the HTML dump too, of course.
I'm fine with a C++ XS version; the big issue is avoiding repeated sub calls so
the wrappers should be at the highest level. In other words, converting
writeI32() to a C++ XS call won't help you because pack() is already very fast.
I don't know the implementation details so please let me know if I should look
at something in particular to contribute better.
Thanks for your help.
> performance improvements for Perl
> ---------------------------------
>
> Key: THRIFT-775
> URL: https://issues.apache.org/jira/browse/THRIFT-775
> Project: Thrift
> Issue Type: Improvement
> Components: Compiler (Perl), Library (Perl)
> Reporter: Ted Zlatanov
> Attachments: nytprof.out.aa, nytprof.out.ab, tester.pl
>
>
> The Perl libraries are 5-20x slower than the Java libraries for the same
> data. Possible performance boosters:
> - make fewer OO method calls
> - build and use monolithic pack/unpack templates when possible
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
