Re: issue 3342

[Edmund Wong <ed...@kdtc.net>]

Stefan Sperling wrote:
> On Tue, Jun 09, 2009 at 10:32:47AM +0800, Ed wrote:
>> Hi Stefan,
>>
>> I've read the description countless number of times and think
>> I need to bounce some ideas.
> 
> Hey Ed,
> 
> I am Cc'ing dev@ in case someone finds errors in my advice below
> or for other reasons disagrees with the approch I'm suggesting.

Thanks Stefan!  This issue is a lot more involved than I thought.
If this is called 'bite-size', I really can't imagine what
a full fledge issue would look like.  ;)

Seriously though, I very much appreciate the help.

> 
> Maybe you know some of this already, but recall that
> in principle, what is happening here is this:
> 
> There is a callback mechanism Subversion uses in its libraries.
> Any client might have their own way of presenting data to the user.
> The command line client might use printf(), a GUI client like
> TortoiseSVN might display things in a Windows, etc.

It's this callback mechanism that hasn't clicked until now
as I've used it before in a milter I made, which
uses callbacks.


> 
> The libraries don't care how data is displayed. They only
> notify the client "hey, here's some data for you to display
> and you can display it any way you like" by calling a "notification
> function" like notify_func2(). This function is provided by the
> client as part of the client context (ctx), and is implemented
> by each client individually. It's purpose is just to display data
> reported to it by the client library.

That clears up the picture quite a lot.  Sure, I still need
some time to digest it and the code, but it definitely helps
me understand what really is going on.  Which is why I
asked the 'take-issue' question.  This is going to take a bit
longer.  I do appreciate the patience.



Edmund

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2360777