This is an old revision of the document!
This idea involves logging (tracking) downloads, namely from here:
The code should:
- Never interfere with ability to download
- Be efficient
- Track information like for example: file name, timestamp of download, file size, manual build date, ?user info?, ...
- Consider being used for other parts of the site, like php source code downloads (ex: php-5.x.x.tar.gz)
- Consider being used for other sites, like snaps.php.net
- CVS Module: phpweb (the www.php.net website)
- Helper functions spread around elsewhere
- CVS Module: phpdoc (the sources to the php manual)
- None needed.
- CVS Module: php-master-web (the master.php.net website)
- Very old, not used: http://cvs.php.net/viewvc.cgi/php-master-web/log_download.php
- CVS Module: phd (the documentation build system)
- None needed. But, it might be adjusted to, for example, create unique ids for each build
Long ago we tracked downloads but it was disabled [in 2004] for a few reasons:
- One day the master server went down, and this stopped the downloads from working
- The data wasn't really used
- Nobody felt the need to work on it
How exactly this will work is unknown. A few questions needing answers:
- How will the data be stored? (Note: write heavy, few reads)
- Where will the data be stored? (Master?)
- Is this useful data? (worth the performance hit?)
- How will the data be compiled and presented?
- Each mirror should behave the same, and work without too much trouble... how?
People who are working on it:
- Probably nobody... you?
It would be helpful to determine and document exactly how our download system works. For example, a 404 handler sees /get/ then ...