[PATCH] Allow adding usernames to a ticket CC list

[Mitar]

Allow adding usernames to a ticket CC list instead of e-mail addresses for users which can only select checkbox to add themselves to the CC list (and thus cannot control content). For our setup it would be enough to just add a project-wide configuration switch which would tell Trac whether to prefer usernames or e-mail addresses.

[anonymous]

This already seems to be working in trac at ​http://sourceforge.net/apps/trac shows it as:

Add to Cc: myusername [ ]
and
Remove from Cc: myusername [ ]
when editing a ticket.

Not sure how to configure it though.

stefan

[mitar]

That is so only because your username does not have e-mail address set up in Trac. In _toggle_cc method there is this code:

entries = []
email = req.session.get('email', '').strip()
if email:
    entries.append(email)
if req.authname != 'anonymous':
    entries.append(req.authname)

...

if remove:
    action, entry = ('remove', remove[0])
elif add:
    action, entry = ('add', add[0])
return (action, entry, cc_list)

So it uses e-mail address if there is one or username. If I change last lines to:

if remove:
    action, entry = ('remove', remove[0])
elif add:
    action, entry = ('add', add[-1])
return (action, entry, cc_list)

it prefers username.

[cboos <cboos@…>]

Patch welcomed.

[anonymous]

Should it ever prefer email address? Usernames are more robust (that is, less likely to change). Plus, adding email addresses means more places for spammers to scrape information from. Also, querying for tickets a user is CC'd on becomes harder, because they might have to search for multiple values.

[mitar]

I agree. Especially because Trac does not support multiple owners of the ticket. So if multiple persons are working on a ticket they are often listed in CC field. So usernames are also preferred in this case.

[mitar]

Patch for this one-line change?

[exarkun@…]

Any hints on where the unit tests for the cc manipulation code are?

[Christian Boos]

Replying to anonymous:

Should it ever prefer email address? Usernames are more robust (that is, less likely to change)

The idea was that you should better grab an e-mail address when you have one, session data is purged after a while.

In later iterations of the notification subsystem, if the e-mail addresses are stored differently, maybe this could be changed.

[Mitar]

Replying to cboos:

session data is purged after a while.

Even for registered users? Is there a way to disable/configure this?

[Christian Boos]

Replying to Mitar:

Replying to cboos:

session data is purged after a while.

Even for registered users?

No, you're right, the session is only purged for non authenticated users. So we could indeed already prefer the user name for authenticated users.

[Mitar]

I am using on a production Trac system above changes and it works well. So only that small change is necessary to prefer usernames before e-mail address. But probably this should be configurable so there is need for some additional code.

[Mitar]

If it is not visible: it is just necessary to change 0 index to -1 index.

[bill_b <bill@…>]

Any chance of this getting into the system soon? Thanks.

[Mitar]

I made a ​plugin for this.

[bill_b <bill@…>]

Replying to Mitar:

I made a ​plugin for this.

Thanks - especially for the quick response. That seems to work well. This does seem like a simple enough and desirable feature to have integrated in, but this works for now.

[holger-juergs@…]

using the following PATCH of web_ui.py (under 0.12) solves the problem for me with permission of ​PrivateTicketsPlugin (​#th11862)

web_ui.py

@@ -647 +647 @@
         of CC status for this user relative to the current `cc_list`."""
         entries = []
         email = req.session.get('email', '').strip()
-        if email:
-            entries.append(email)
+#        if email:
+#            entries.append(email)
         if req.authname != 'anonymous':
             entries.append(req.authname)
+        elif email:
+            entries.append(email)
         else:
             author = get_reporter_id(req, 'author').strip()
             if author and author != 'anonymous':

perhaps there will be some side-effects…