Re: [ jEdit-devel ] [ jedit-Patches-1672086 ] Cleanup for Clear Undo on Save

[Kazutoshi S. <k_s...@f2...>]

Hi Matthieu,

>>Comment By: Matthieu Casanova (kpouer)
(snip)
> Hi, I am late but what was the reason of this patch ?

Sorry, there was a discussion in some private mails sent only to Joseph
and Marcelo. I attached the summary of it at the bottom of this mail.
I wasn't sure that it should be on the list at that time. But I'm now
absolutely sure it should be.

> It introduce a regression since the UndoManager must not depend on jEdit
> class.
> Can it be rollbacked ?

I see the problem. But I don't want rollbacking it for reasons said in
the mailing. I'll make a patch to solve it, do some test, and commit.

Could you please wait for a while? It will one or a few days.



The mailing started at 2007/03/01:

Kazutoshi Satoda wrote:
> va...@us... wrote:
>> Revision: 9037
>>           http://svn.sourceforge.net/jedit/?rev=9037&view=rev
>> Author:   vanza
>> Date:     2007-02-27 21:18:25 -0800 (Tue, 27 Feb 2007)
>> 
>> Log Message:
>> -----------
>> SF Patch #1666690: No Undo past file save option
> (snip)
>> Modified: jEdit/trunk/org/gjt/sp/jedit/bufferio/BufferSaveRequest.java
>> ===================================================================
>> --- jEdit/trunk/org/gjt/sp/jedit/bufferio/BufferSaveRequest.java	2007-02-27 14:22:46 UTC (rev 9036)
>> +++ jEdit/trunk/org/gjt/sp/jedit/bufferio/BufferSaveRequest.java	2007-02-28 05:18:25 UTC (rev 9037)
> (snip)
>> @@ -137,6 +137,9 @@
>>  				finally
>>  				{
>>  					buffer.readUnlock();
>> +					if(jEdit.getBooleanProperty("resetUndoOnSave")){
>> +						buffer.getUndoManager().clear();
>> +					}
>>  				}
>>  			}
>>  			finally
> (snip)
> 
> This code invokes buffer.getUndoManager().clear() even if the save request
> failed (with something like a low level I/O error, an encoding error, ...).
> User will get nothing but loses undo. Is this intended?
> 
> Should I use devel-list or commits-list for questions like this?

Joseph Erickson wrote:
> Completely unintended.  It makes sense now that you say it though.
> 
> Where in the code is a successful save certain?  Right after
> write(buffer,out); in the try block?  Is it later on when the markers
> file is cleaned up?  When vfs.saveComplete is called?  There's a lot of
> try-catch-finallys in there and it's hard for me to tell where things
> are happening.
> 
> Let me know and I'll submit another patch.

Kazutoshi Satoda wrote:
> I propose to place it in Buffer.finishSaving(), next to
> undoMgr.bufferSaved() in it. Or, placing in UndoManager.bufferSaved()
> may be more correct. In latter case, you should take care of that
> UndoManager.bufferSaved() is also called from Buffer.setDirty().
> In both way, new method getUndoManager() will be no longer needed.

Joseph Erickson wrote:
> Thank you for your help.  I refactored bufferSaved() to handle this and
> moved the dirty flag handling code into a new method.
> 
> http://sourceforge.net/tracker/index.php?func=detail&aid=1672086&group_id=588&atid=300588

-- 
k_satoda