9

Please forgive me that this has been discussed multiple times on meta.SO. I'm bringing it up here for two reasons: 1. The problem doesn't seem to be resolved, 2. I'm not sure if the tex.sx community is aware of it.

It doesn't happen too often on tex.sx, but today it did happen: TH saved an edit of this question with the time stamp 2011-01-23 02:11:49; then lockstep saved an edit with the time stamp 2011-01-23 02:12:01, so 12 seconds later. Both edits were to mark a code sample; TH's edit actually involved one more change than lockstep's, but this got overwritten by lockstep's edit.

To make things worse, I made another edit (hours later). Why worse?  There's a little known (I believe) feature on stackexchange that frequent edits of a post make it CW, namely 5 edits by different users or 10 edits by the OP.

So having multiple users edit one and the same post has two downsides: 1. Only the edits of the user who save his edits last take effect; thus the effort of the other users is wasted. 2 There's the danger of making a question CW shortly after it has been posted (so far not a great danger on tex.sx).

There is already some warning implemented, but this doesn't seem to suffice. I've got two completely different suggestions:

  1. When I hit the "edit" button shortly (let's say <5 minutes) after another user, the system shows a message like

    Another user may currently be editing this post. Do you want to proceed?

    (Very similar to this proposal; should the name of the other user(s) be given?).

  2. When I hit the "Save Edits" button, the system shows a message like

    Another user has edited this post while you were editing. If you proceed to save your edits, his edits will be lost.

    (This could still be improved a lot, see e.g. this old suggestion, but my present proposal would be simpler to implement.)

Maybe my points 1. and 2. could/should both be implemented? (Of course I can't judge if the implementation is possible at all in the current framework.) I'd like to know what others think.

7

This already exists -- the system pings the server every 45 seconds once an edit is initiated -- if another revision happens while you are editing, you will be notified of it.

Will think about #2 but I have serious concerns about good edits being thrown away just because they might conflict.

UPDATE: (A version of) #2 has now been implemented. Citing Jeff from meta.SO (Notifications for concurrent changes when editing posts, 2011-03-30):

Additionally, we now prevent edit collisions on save.

When you click "Save Edits", if the post changed while you were editing, we now return:

{username} edited {tags / body / title} of this post; try refreshing this post and editing again.

It is specific to the area in question, so one user can edit body and you can edit tags without conflicting. (Or title, etc). It's only when you both edit the same field at the same time this error is returned on save.

  • Sorry for the -1, but did you read that I explicitly wrote "There is already some warning implemented, but this doesn't seem to suffice"? And that I made a feature request to implement something in addition? I find it rather strange that you say this is "status-completed". It's not! – Hendrik Vogt Jan 23 '11 at 11:21
  • @hendrik it is completed from my perspective; if you want alerts every 5 seconds (eg between 02:11:49 and 02:12:01), you need to look elsewhere on the internet. If your opinion is "this doesn't seem to suffice", well, I beg to differ. – Jeff Atwood Jan 23 '11 at 11:48
  • @Jeff: Thanks a lot for your reply. Note that I didn't ask the 45 seconds period to be reduced to 5 seconds; I asked for something in addition. If my two suggestions both are impossible (or too hard) to implement, or if they both just don't make sense, then I'd be very happy if you could give me a short explanation. – Hendrik Vogt Jan 23 '11 at 12:08
  • 2
    @hendrik I will think about #2 but it has not been a large problem to date; I don't like the idea that a good edit might be thrown away just "in case" it conflicts. – Jeff Atwood Jan 23 '11 at 12:40
  • @Jeff: Thanks. The "thrown away" business is adressed here, but I can imagine that this would take some effort to implement. If you find time, I'd be really happy to learn why you don't like #1. (I'm clearly not qualified to judge what is good/bad or possible/impossible!) Moreover, could you please unlock the post? I won't edit it again, but people could leave answers. (Incidentally, at the moment it looks as if "status-deferred" would suit better than "status-completed".) – Hendrik Vogt Jan 23 '11 at 16:35
  • 1
    @Jeff: +1 for "will think about #2". Just thinking, too: Good edits will be lost both with and without #2. The difference: Without #2 it can happen that an edit gets overwritten without anyone noticing; with #2 it can happen that the edit doesn't get posted at all, but that's the decision of the user who edits. By the way, for this post I'd have appreciated a feature like #1 or #2 a lot. – Hendrik Vogt Jan 24 '11 at 10:43
  • @Jeff: Ah, my last comment on meta.SO should have gone here: Thanks for implementing (a version of) #2! Terrific!! Now I'd also say this is "status-completed". I'll take the liberty and edit your answer. – Hendrik Vogt Mar 30 '11 at 13:24

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .