= Description =

I get subscription mails for some pages on http://moinmoin.wikiwikiweb.de (e.g. WikiSandBox) that are not in my subscription list (user RobertSeeger).

== Steps to reproduce ==

Difficult to say, it just happens that I get subscription mails from wiki@wikiwikiweb.de about pages like {{{http://moinmoin.wikiwikiweb.de/WikiSandBox}}} that I do not think I've subscribed to.

== Example ==
##Add URL that show the bug, screenshot or test wiki markup that fail...
Here is the subscription list from my {{{UserPreferences}}} page:

{{{
AccessControlList
MoinMoinBugs/AccessControlListAddLine
ApiChanges
MoinDev/MoinMoinLogo
.*Market.*
FeatureRequests/EditorInUpdateNotificationSubject
MoinMoinPatch
FormatterRefactoring
MoinInternationalConference
FrontPage
MoinMoinBugs/FreeParentLinksAreBroken
MoinMoinBugs/SubscriptionMailLooksBadInOutlook
MicrosoftWordConverter
}}}

##URL:

##attachment:screenshot.png

##{{{
##Example wiki markup that fail
##}}}

== Details ==

## If you got a traceback, please save the traceback page as html and attach here:
## attachment:traceback.html
This Wiki

== Workaround ==
## How to deal with the bug until it is fixed
Just ignore the unwanted subscription?

= Discussion =

Two questions:
 * Can you reproduce it? try to edit WikiSandBox in another browser, as anonymous user. Do you get a notification for every edit?
 * Are you subscribed to "Trivial edits"? try with and without that setting.

Pointless! The wiki shows him as a subscriber ...

 -- RobertSeeger <<DateTime(2005-08-31T10:15:59Z)>> sorry, should have checked this first myself! After removing the regex line {{{.*Market.*}}} I am no longer subscribed to WikiSandBox. So the problem must be something about regex search in subscriptions. However using this regex in a title search returned exactly what I'd expected, so I was not suspicious. Am I doing something wrong?
 No.

So the regex is matched incorrectly.

 -- RobertSeeger <<DateTime(2005-08-31T10:35:23Z)>> seems that I'm subscribed to all pages that have {{{CategoryMarket}}} (including WikiSandBox and FeatureRequests, which do not appear in the title match for {{{.*Market.*}}}in their text - is there any special rule on categories? I thought that subscriptions only consider page titles, not content?
  Yes, they should just consider the title.

Hmm, what happens with empty lines...?

 * userform.decodePageList remove empty lines
 * user.encodeList remove empty lines
 * user.decodeList remove empty lines

So it should not be empty lines problem.

/!\ This is not a bug, but a feature. If you subscribe to a Category page, you will get notifed when a page "contained" in that Category changes. Your regex `.*Market.*` matches Category``Market and when someone puts Category``Market on WikiSandBox, you get notified of WikiSandBox changes...

-- RobertSeeger <<DateTime(2005-09-26T09:05:49Z)>> If this is really a feature (and I'd like you to reconsider this), then it is not documented in HelpOnUserPreferences, as this states:
 "If you are familiar with '''regular expressions''', you may enter a regex expression to match the pages names of interest (.* matches all page names)."
It does not say anything about the regex matching the '''content''' or category of a page. If you think of the consequence this would mean that you can't subscribe to any page name regex that ''may'' match any category page name or full text anywhere without the risk of getting plenty of unwanted mail. Just think of those damned spammers that frequently try to pollute this and other wikis.
Subscription to full text may have its benefits, but maybe you could make this an optional feature that only works if you prefix the regex with something like FULLTEXT or CATEGORY?

If you find other people editing this page to avoid it showing up in unexpected places then maybe you see that this "feature" has its negative effects. There should be a clear separation between title search, fulltext search and - maybe - category search. In the future I would recommend to define things like category, ACL etc. as "properties" of a page that are not part of the page text.
----
I agree...  Category subscriptions are powerful, but there should be ''some'' special handling, such as preventing pattern matches from activating the feature.  If you want to subscribe to all pages in a category, I think it's reasonable to require that you place the full name of the category in your subscription list.  If you want to subscribe to multiple categories that happen to share a similar naming pattern, just put each category name in the list individually.

I have also been having great trouble finding information on category use, since almost every page has the word "category" on it somewhere.  It appears that there is nothing in the help explaining the use of category subscriptions (or even mentioning that the feature exists), but possibly it's just that I haven't found it yet because of the problem I just mentioned.  But '''something''' needs to be done IMO to differentiat categories from regular text.  -- SteveDavison <<DateTime(2006-03-27T21:09:33Z)>>

= Plan =
## This part is for Moin``Moin developers:

 * Priority:
 * Assigned to:
 * Status: 

----
## If you are a moin core developer, replace the category to Category* in these cases:
## Category MoinMoinNoBug - if this is not a bug.
## Category MoinMoinBugConfirmed - if you can confirm the bug on current code.
## Category MoinMoinBugFixed - after the bug is fixed in current code.
CategoryMoinMoinBug