-
Notifications
You must be signed in to change notification settings - Fork 2
Aliases location #17
Comments
Alias with wildcard (
Looks a good idea.
I guess the reason as below:
For example, when the disk crashed and alias files were lost, list aliases can be regenerated by
In a sense, these alias files are not mandatory. Because their contents are differ from MTA by MTA, and there are several ways for single MTA (e.g. P, S, O). I feel it is possible that these may be provided as sample files, not as working configuration.
|
Thanks for the detailed explanations. |
Hi Soji,
Currently the sympa aliases are located in 3 places:
First, I'm not sure the tiny difference between the sendmail and postfix aliases file really needs to split them in two. Probably, merging them and adding a comment would be enough.
Also, the rpm would need to ship empty /etc/sympa/aliases.sympa.sendmail.db and /etc/sympa/aliases.sympa.postfix.db owned by root:root and mode 644 so the newaliases command would work out of the box.
Second, I'm not sure to understand why the split between the robot and list aliases is done. My uneducated opinion is it adds yet another complexity to the setup, one needs to define 2 aliases files in the MTA.
Also, same issue as above with the missing 644 root:root /var/lib/sympa/sympa_aliases.db
Both files would need to be created in %post, only if they don't exist.
I'd rather have them directly shipped in the rpm and marked as %config(noreplace), but then rpmlint would rightly complain about config files not in /etc.
I would like to suggest to gather them all in /etc/sympa/aliases, for ease of setup and packaging.
I have a branch which implements these changes (although the merge of sendmail/postfix aliases and .db creation is in one commit which would be better split in two).
https://github.com/xavierba/sympa-rpm/tree/aliases
In case we follow that path, some additional commits will be needed to update the doc and ensure smooth migration of the aliases, but I'd rather add them once we have an agreement on the way forward.
I'm probably missing some background and all my changes might not make sense. You know the codebase much better than I ever will, so I'd really like to have your educated opinion.
I think this is the last thing to address before the package is ready for a review.
Regards,
Xavier
The text was updated successfully, but these errors were encountered: