[pmwiki] [Pmwiki-users] CVS features was: discussion on linebreaks

[Bernhard.Weichel@t-online.de]

Crisses wrote:
>> Page locking is really hard to handle in a system which is not
>> transaction
>> based. An easy think could be a warning that someone else has started
>> editing a page withing the recent 10 Minutes. Or a warning that
>> someone else has an open editing session. It could even go that far,
>> that PmWiki
>> checks, if the page was changed since the beginning of the editing
>> session.
>
> I think the least obtrusive way to handle it is a version/timestamp
> check, as Patrick is saying.

Yes, but this shows the conflict when the change is submitted. This might be
ok.


> Is it possible that when the form submission has a different version
> than the latest version of the document, a new query page is
> implimented so that the user is given two form boxes:  "This is the
> material you are attempting to submit" and  "this is the latest
> version of the page in question" -- this way one can cut & paste from
> one form text entry box to another without the loss of the newly
> input data?

this is the approach of Wikipedia:

1. Textbox: most recent version only this one has the submit button.
2. rendered difference
2. the current submission from which the changes can be cut and pasted
   to the most recent version.

> Each box can have a "submit this version" button I guess...making it
> easy to use either window to resubmit data, since it might be easier
> to edit from either window.

IMHO submit is sufficient in the most recent version.

> This page could include an "Use automatic merge feature" button, ,as
> well.

This would be an extra plus. OTH I do not really trust automatic merge.
While usin CVS, up to now any automatic merge has to be checkt very
carefully.

But if we want the merge feature, then it should be such that it merges
the current submission into the most recent version, displays the result
in the top edit box and also shows the difference between the merged
version an the submitted version.

But again, I do not trust mergers.

> This issue is a little complex, unfortunately, but I don't see why it
> requires logins and sessions to do it.  And if you're not locking
> people out of editing pages that others are editing (an idea I don't
> like either), you're going to hit the same problems whether or not
> people are logged in.

This is basically true.

> I don't like the "X is editing this page, submit anyway?" solution.
> Perhaps if it were a multiple choice question such as
>
> Alert!  This page has a more recent submission than you just edited
> from.
> Would you like to:
> [] See the latest page in a new window?
> [] Merge your information into the new document automatically?
> [] Re-edit the page from the newest version? (this will lose your
> recent editing work)
> [] See your submission and the new data side-by-side in an editing
> window?

This would probably be little complicated. The Wikipedia approach described
abov is fine.

Bernhard