Optimized Tokenizer.php #130
Conversation
MichaelRoosz
force-pushed
the
optimized_tokenizer
branch
2 times, most recently
from
b78d7bb
to
7493665
Compare
MichaelRoosz
force-pushed
the
optimized_tokenizer
branch
from
7493665
to
2414194
Compare
Allow entity references containing numbers
…lotz/html5-php into MichaelHeerklotz-optimized_tokenizer
|
Hi, thanks for your PR. Looks good.. give me some time to check it properly |
|
Any timeline on this? I know things like this can be quite intricate, but I'm about to release a new major version of wp-Typography and any speed increase in the tokenizer would be very, very welcome. |
|
Most probably will have a look at it on Monday
…On 25 Aug 2017 11:35, "Der Mundschenk & Compagnie" ***@***.***> wrote:
Any timeline on this? I know things like this can be quite intricate, but
I'm about to release a new major version of wp-Typography and any speed
increase in the tokenizer would be very, very welcome.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#130 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAvaJ1a8MBrmgIM8KZ7vugcFTR4iw2CAks5sbpV1gaJpZM4Ogewd>
.
|
There was a problem hiding this comment.
On the whole I can see how this would be a speedup and the changes appear to make sense.
@goetas Being that it's changing a protected API (that impacts classes that would extent this) is this API change considered breaking (from a semantic version perspective)? This might be better on the 3.x branch and that be released with some more changes.
| do { | ||
| $p = $this->scanner->position(); | ||
| //$p = $this->scanner->position(); |
There was a problem hiding this comment.
I wonder why we get the position and don't use it? In any case, these can likely be removed rather than commented out.
| dist: precise | ||
| fast_finish: true | ||
|
|
||
| dist: trusty |
There was a problem hiding this comment.
I'm likely missing something but, why are precise and trusty being specified?
There was a problem hiding this comment.
trusty does not support php 5.3, so precise is necessary.
trusty was not the default build platform on travis ci at the time of the commit
|
closed by #135 |
I am working on a project where we parse several thousand HTML5 pages with this library.
When profiling the code with xdebug, I noticed extreme amounts of calls to the scanner's .current() and .next() functions. Here is my attempt to optimize the tokenizer code.
Run time comparison:
Original Code: 281s
Only with the optimized quotedAttributeValue() function: 242s
With all optimizations: 226s
-> Saves about ~60s / 20%