Actually, the op did give you enough information, although I think he also made a common mistake about regexp. However, I'll try to fill in the gaps.
When setting up Drupal for multiple sites on one server, you are advised to not use the default site installed which becomes the "template" for all others. That means you wind up making entries in /etc/drupal/6/sites/mydomain.com There are two files of interest in there: settings.php and (if you create it) baseurl.php. The baseurl is used to give the particular site its base domain name (like http://mydomain.com). The script in question is cron.sh: /usr/share/drupal6/scripts/cron.sh. It attempts to use curl to load http://{yoursite}/cron.php. If it can find the basename, that's good. If it can't find the basename it uses http://localhost which in a multi-site set up is probably not present and therefore you get a "cron failure" e-mail. Even if it is present, I am not sure running cron.sh for one installation in a multi- site install is sufficient (might be, but I doubt it). Now, why is this a problem? Well it turns out it isn't exactly, although it could be improved. You would THINK that since $ is a regexp meta character (match end of line) that if you use it somewhere in a regexp it should be escaped. I would just because it makes sense. However, at least the version of grep shipped with the system understands that a $ followed by other characters must just be a dollar sign. So the script is "correct" but easy to confuse that it is not correct. HOWEVER, there is a common case where the script will fail. It assumes that the base_url file (or settings.php) will use single quotes around the URL. This isn't necessary as you could use double quotes and get this sort of failure. You also get the failure anywhere you might have otherwise harmless whitespace So you could write a big nice regexp to handle it. Something like: ^[ \t]*\$base_url[ \t]*=[ \t*][\'\"] Or even: ^[ \t]*\$base_url Would probably be sufficient. However, if you do anything "fancy" that still could break (e.g., look up a site name in a database). Somewhat less efficient, but you could in fact run a php script (would require php-cli though which is currently not a dependency, I think) and have it include the other files and then echo $base_url. That would be the gold standard. So to summarize: 1) OP's bug is not really a bug because the $ metacharacter seems to be position sensitive and is treated as a normal character if not at the end of string. 2) However, the regexp is bugged because it requires you to adhere to a single format for the base_url entry in either configuration file it checks. 3) Suggest either fix the regexp as outlined or switch to using php to interpret the variables. -- Regular Expression error in cron.sh https://bugs.launchpad.net/bugs/314815 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
