Re: Question collaborative editing - Wikipedia reference

[joakim]

Qiantan Hong <qhong@mit.edu> writes:

> Here is the current version, I feel like a relatively complete 
> set of essential features is implemented now.
> Changes:
> - Unicode support
> - fixed cursor behavior
> - M-x crdt-list-users displays a list of active users. Press enter
> to go to the position of that user’s cursor

Thanks for this work, I will try it within our team during one of our
online sessions.

I tried it briefly myself between 2 emacsen and it seemed to work
nicely.

/Joakim

>
>
>
> It seems that locally I didn’t reproduce yanking issue and multiple 
> crdt-test-client issue — it just open another same document normally,
> which is not the best behavior arguably but I didn’t get any error.
>
> Also in terms of further development, yes a repo will be very helpful
> if anyone else is interested in contributing. I used GitHub before
> but I’m not sure if it feels comfortable for everyone — using it only
> as a file hosting service makes me feel better but it does run whatever
> software MS wants to run under the hood…
>
>
>> On Oct 8, 2020, at 10:42 PM, Yuan Fu <casouri@gmail.com> wrote:
>> 
>> 
>> 
>>> On Oct 8, 2020, at 10:12 PM, Qiantan Hong <qhong@mit.edu> wrote:
>>> 
>>>> Another question: According to the algorithm, what happens when both 
>>>> cursor are at the same place and one inserts text? Is the other cursor 
>>>> pushed forward or not?
>>> Currently the other cursor will not be pushed forward. Any suggestion for 
>>> “better” behavior?
>>> Also note that the cursor-treating part of version on this mailing list is 
>>> not very polished.
>>> I’ve made some fixes and improvements on my local version, will post 
>>> shortly.
>>> The new version will also contain Unicode support and active user list.
>> 
>> Pushing forward is more intuitive IMO. I asked because I see some small 
>> problems with the cursor overlay but I’m sure you’ve fixed them now.
>> 
>>> 
>>>> Very cool! If I try to run crdt-test-client when already connected, Emacs 
>>>> creates a new buffer and compains "Error in post-command-hook 
>>>> (crdt--post-command): (wrong-type-argument stringp nil)”. Any idea why?
>>> Thanks for the report! I haven’t consider this case. What do you think is a 
>>> reasonable
>>> behavior? Should it make a new connection?
>> 
>> It probably doesn’t make much sense to have multiple connections on a single 
>> Emacs instance to the same document. I’d just stop and complain to the user.
>> 
>> Yuan
>
-- 
Joakim Verona
joakim@verona.se