mirror of
https://github.com/lubuntu-team/lubuntu.me.git
synced 2025-02-24 16:51:07 +00:00
56 lines
3.7 KiB
Markdown
56 lines
3.7 KiB
Markdown
Changelog
|
|
-------
|
|
|
|
### Version 2.0.0 - June 23, 2014
|
|
- #25 for caching error
|
|
- #14 reducing memory consumption slightly
|
|
- #22 use the "normal" version instead of "full" version
|
|
- #15 general improvements
|
|
- #13 code formatting improvements
|
|
- #12 allow proxy settings via fsockopen with context
|
|
- #9 default timeout settings fix
|
|
- #11, #27 improved tests
|
|
|
|
### Version 2.0.0-beta - August 25, 2013
|
|
- Added a new method to retrieve the source file version `$browscap->getSourceVersion()`.
|
|
- Added a safety feature to regenerate the cache file always when `Browscap::CACHE_FILE_VERSION` changes.
|
|
- Updated source file download URLs to new temporary URLs.
|
|
- Added new lines `\n` to the cache files for readability.
|
|
- Default download URL is changed so it will get and parse the full file instead of 'standard'.
|
|
[ua-speed-tests](https://github.com/quentin389/ua-speed-tests) shows that there is only a small performance difference between
|
|
using those two versions.
|
|
- Performance upgrades (see [ua-speed-tests](https://github.com/quentin389/ua-speed-tests) for performance tests):
|
|
* **5 times faster** for real user agents, with opcache on
|
|
* **11 times faster** for user agents that do not match anything, with opcache on
|
|
* **3 times faster** for real user agents, without using opcache
|
|
* **5 times faster** for user agents that do not match anything, without using opcache
|
|
* Regular expression pattern matches are being grouped by version numbers. The matches are performed
|
|
in two stages. 1 - standard regular expression match with numbers that differ across source file
|
|
patterns replaced with single character wildcard match. 2 - a check is performed on found numeric
|
|
values to see if any of the grouped values are an exact match. If not, then the searching process resumes.
|
|
This is the main source of the speed optimization. It greatly reduces the source file size and
|
|
greatly increases matching performance.
|
|
* Data that is not required to perform matches or return results was removed from the cache files.
|
|
That includes the source file match strings, which can be recreated from the regex ones, and a large set
|
|
of browser name entries which were never used because they had parents. Decreasing the cache file size
|
|
is very important for when you don't use any PHP opcache, because loading large data structures into
|
|
PHP takes a very long time.
|
|
* Arrays that are not used in `foreach` loops were serialized in the cache file. This also decreases the time
|
|
it takes to load the cache file when not using opcache. It's generally a very bad idea to load large arrays
|
|
with subarrays into PHP. Serializing does a great job when optimizing performance.
|
|
* The above changes address performance issues brought up in https://github.com/GaretJax/phpbrowscap/issues/26
|
|
- Bug fixes:
|
|
* Fixed https://github.com/GaretJax/phpbrowscap/issues/35
|
|
* Fixed https://github.com/GaretJax/phpbrowscap/issues/34
|
|
* Fixed https://github.com/GaretJax/phpbrowscap/issues/33
|
|
* Fixed https://github.com/GaretJax/phpbrowscap/issues/32
|
|
* Bug https://github.com/GaretJax/phpbrowscap/issues/17 is resolved, although that was fixed even before.
|
|
* Merged https://github.com/GaretJax/phpbrowscap/pull/25 - those are mainly comment changes but there are also two fixes
|
|
for `$browscap->clearProxySettings()` method, which did not work properly when an optional `$wrapper` parameter was passed.
|
|
- Added a new testing class that compares result of `Browscap` to `get_browser()` for as many browsers as possible
|
|
and checks if there are any differences in parsing. It also compares the parsing speed (in a simplistic way,
|
|
more advanced tests are available at https://github.com/quentin389/ua-speed-tests).
|
|
|
|
### Version 1.0.0 - July 22, 2012
|
|
- Initial version
|