Allow different content marker types

Stefan, markers are not necesserily inline. At one of my sites, we use a right floated block element to indicate that something is new. So this kind of help text will not work... The marker however can contain some help in itself in a title="" attribute, so that those hovering over it with their mouse can get an idea on what it is.

Suprise :) In real incremental patching sense, here is a patch to introduce the different new and updated makers. It operates with passing on the info to theme('mark'), and then letting it decide on what to do with it. This also enables people to do 'read' markers, if so desired (displaying different icons for read, uread and updated content).

Renamed the 'new' marker to 'content', so that it is not misleading, and introduced some constants, to help along the way. 'node' might be a better name instead of 'content', but it is just a search and replace in the patch, so Dries can fix it up, if needed. Also introduced the unkown maker for anonymous users. The practice before was that by setting the last view time to time(), markers are always displayed as 'read' for anonymous users. But with the different icons possibility in mind, it is important to distinguish the case when we cannot decide on whether a content is new/updated or not, because we have no data. This is why the unkown marker is introduced.

I admit I have not tested the patch, but it should work well, theoretically. Kept the original interface behaviour, which does not expose much from the power of the markers, but preserves the currently expected output.

Wups, forgot to adjust one usage of node_is_new(). Plus renamed it to node_marker(), since it is not anymore only about checking on whether it is new or not...

Markers are only important for logged in users, yes. We need to somehow omit the markers. We can check for the user in theme_mark(), but we need to check for the user in node_marker() anyway, since otherwise we would do an unnecessery database lookup for each marker. So we can return MARKER_READ in node_marker() and omit the marker regardless of the $marker if the $type is 'new' and the user is not logged in. So we need to check for the user in the theme function too...

BTW the terminology I introduced seems to be inconsitent: node_marker() vs. theme_mark(). It might be desireable to rename the theme function, or use MARK_ constants and node_mark(). Pick the way.

Since noone hinted on a way to pick, I needed to decide myself :) Gone is the unkown marker, and now it is the theme's task to not display content markers for anonymous users (while still keeping the required marker for them). The contants and functions were renamed to use 'mark' instead of 'marker' in line with the existing theme_mark() function. Still the same functionality, but with less code compared with the previous patch.

Updated patch, which links to theme_mark() and node_mark(). Otherwise I think the names of the constants really tell everything.

Updated is not unread. Also there could be 'read' marks. Your email client has marks for read and unread messages doesn't it? Drupal will allow this too. If we rename 'content' to 'unread', then the unread marker would be able to signify that something is changed (while it was already read) and that it was read (which is odd for an unread mark you must admit :).

OK, Dries is right again :) Updated patch with markers separated. The form required marker has it's own CSS styling now, and only the content markers are printed by theme_mark(). Also changed the comment code a bit to delegate marker display to the theme function. This does add an extra space after the comment subject if a marker is not printed, so it might not be desirable. Note though that there has always been these kind of extra spaces after the node titles in tracker and node modules on the user interface, and after comment titles on the admin interface. So I don't think this is becoming a problem.

I hope this patch is going to be fine. :)

Tangent, you do suggested 'unread' as a replacement to 'content' above. Or at least I tried to read your comment multiple times, and understood this suggestion. The latest patch has no 'content' in it, so your remarks are not applicable anymore.

If Dries wishes to replace MARK_NEW with MARK_UNREAD, then it is fine with me, it is just a simple search and replace in the patchfile. I think NEW vs. UPDATED is better distinction then UNREAD vs. UPDATED, but the way I originally choose was not what is reflected in the current patcfile, and it changed for the better :)

Tangent, if you now a way to distinguish between the appended/extended state and the updated/modified state, then feel free to work on the patch more. I have no idea on how one would be able to distinguish between these two. If I add a paragraph to a node, is it a change or an extension of the content? If I modify a sentence and add three words in? Is it a modification or an extension?

1. Tangent you should interpret new "as being created with a certain time frame" AND also being unread. Once a node becomes too old, it is not marked NEW anymore, regardless of one reading it or not. This behaviour was Drupal's own for quite a long time now.

2. No, I have not changed back the label to 'new'! The 'new' label was always there. You misunderstood something. Dries was also confused :)

3. We are not just modifying a theme function, but also introducing a node function, which does provide the node state. It is the task of that function to decide on state. Please look and the explanation carefully. I have described this earlier and the code for the node function is also in the patch. If you would have looked to the patch, you would have not been confused by the meaning of NEW (see point 1 above).