Added user list page at _users

[duairc]

This displays a table of all the users of the site, including the times they last logged in.

By default, the table does not show user's email addresses. I added a config option that enables this which is off by default. When displaying email addresses, it uses pandoc's reference obfuscation.

I also added two fields, both UTCTime, to the User data type: uCreated and uLastSeen. In order for this to be backwards-compatible with existing installs of gitit, I added a hand-written Read instance for User, which can parse the old User type as well as the new one.

[jgm]

What if you stop the gitit instance and start it up again? Is the information about login times persistent? This would presumably require writing to the users database file. I am not wild about the idea of rewriting that file every time anyone makes a change...

[duairc]

The information about login times is persistent; that stuff is in the first commit. What I do is call adjustUser every time a user logs in, which from what I can see, does in rewrite the users file every time it's called. It doesn't do this when a change is made, it does it whenever a user logs in (which is probably just as bad).

[jgm]

Well, my instinct was to avoid touching the login file that frequently,
but on reflection, I'm not sure what would be so bad about it.

+++ Shane [Apr 12 14 12:35 ]:

The information about login times is persistent; that stuff is in the
first commit. What I do is call adjustUser every time a user logs in,
which from what I can see, does in rewrite the users file every time
it's called. It doesn't do this when a change is made, it does it
whenever a user logs in (which is probably just as bad).

—
Reply to this email directly or [1]view it on GitHub.
[3044__eyJzY29wZSI6Ik5ld3NpZXM6QmVhY29uIiwiZXhwaXJlcyI6MTcxMjk1MDU1NSwi
ZGF0YSI6eyJpZCI6MjkxNDgwNzB9fQ==--b8e838672f36e09016fac2522caae004ab5b3
afb.gif]

References

1. Added user list page at _users #411 (comment)

[duairc]

I can understand that instinct. I'm not sure about the specifics, but my worry would be if two users log in at exactly the same time and that somehow some sort of race condition occurs? But seeing as updateGititState uses atomicModifyIORef that should be fine, unless there are two separate gitit instances running at the same time which use the same gitit-users file, but I don't think this a use case we need to worry about supporting.