After a little bit of work, prompted by koobs, I’ve made the following code changes. Lines 8-10 added a time value to the top of the page. This was intended to let the user know when they last refreshed this page. Lines 18-20 let me know how long it took to generate a given page. [...]
It’s one thing to identify a bug. It’s another to clean up the data corrupted by that bug. Fortunately, FreshPorts takes its data from the FreeBSD ports tree. This makes it rather straight forward to clean up most data issues; you just refresh the database from the ports tree. Case in point: a recently found [...]
This just came to my attention. The processing of pkg-descr since we moved to using a jail-based solution is wrong. The function which reads and processes the contents of pkg-descr is not using the jail to read the file. I’ll fix the bug soon, and refresh all commits which touched pkg-descr. Noted by koobs with [...]
koobs mentioned I need to look at etag. See http://redbot.org/?uri=http%3A%2F%2Fwww.freshports.org%2Fsecurity%2Fpy-pycrypto%2F
For some time, the website has had an animated bug. This was originally introduced to the development website as a visual clue for myself. It escaped into the wild through a commit. Today I’m fixing that mistake by making the image a configuration option of the website in question. The diff for this change is:
The problem Sometimes the data in a field changes over time. Or rather, what you expect to appear in the field changes. Case in point, the committer field in a commit email is not always the user id of the committer. Sometimes it contains spurious characters. This is one of the classic problems associated with [...]
Earlier today I run cvsup on the main website and pulled down the code for running make -V in a ‘jailed’ environment. It’s not actually a jail; it’s more of a chroot. You can read up on previous jail posts first, but the idea came from Ryan Steinmetz (zi@). I found the idea incredibly simple [...]
I should have a look at net/svnup and see if it’s better for FreshPorts than svn.
Compare the pages for ruby18 on prod, dev, and beta. It seems they disagree on what versions are affected. I’m not sure if this is a pkg_verison issue or not.
If you’re committing a port, the physical category must be included in CATEGORIES. That is, sysutils/bacula-server/Makefile must have sysutils in CATEGORIES. When people move ports, they often overlook that issue. In a recent commit, 25 ports failed this test. Proposal: add a new sanity test which checks for this situation and alerts the committer.