Thanks. I must admit Xdebug has a pretty strange definition of executable code. For example, variable initializations are not counted.
--- On Sat, 8/1/09, Happy-melon happy-melon@live.com wrote:
From: Happy-melon happy-melon@live.com Subject: Re: [Wikitech-l] Statistics now on MediaWiki page To: wikitech-l@lists.wikimedia.org Date: Saturday, August 1, 2009, 3:44 PM This is something of a hyperbole, it's true; my apologies. Parser.php itself has ~5,200 lines of code (total, including comments); combined with the preprocessors (Preprocessor_DOM.php and Preprocessor_Hash.php, ~1,500 and ~1,600 lines), CoreParserFunctions.php (~650 lines), and the rest of the 'parser' related files in the /parser directory (~300 lines each), you get around 11,000. This is total lines, including comments. ~3,000 executable lines in Parser.php sounds plausible.
--HM
"dan nessett" dnessett@yahoo.com wrote in message news:459215.97119.qm@web32507.mail.mud.yahoo.com...
I am not finished with the analysis (MacGyver) tool,
but I thought I would
put up what I have so far on the MediaWiki site. I
have created a web page
in my user space for the Parser Test code coverage
analysis -
http://www.mediawiki.org/wiki/User:Dnessett/Parser_Tests/Code_Coverage
I would appreciate it if someone familiar with the
parser would at least
glance at the per file statistics for a sanity check.
Some things that
worry me are:
- parserTests seems to visit Special:Nuke. Does this
make sense?
- Only about 72% of Parser.php is exercised. Is this
reasonable?
- Xdebug is reporting that the Parser only has 2975
lines of executable
code. This contrasts to the report by Happy-Mellon
that there is 11,000
lines of code in Parser.php. Are there really that
many non-executable
lines of code in the parser or is Xdebug missing a
whole bunch?
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l