cherry-picking with transplant?

[Mads Kiilerich]

On 03/27/2009 01:47 PM, Greg Ward wrote:
> 2009/3/26 Arne Babenhauserheide<arne_bab at web.de>:
>    
>> Should we open a bug "transplant should use the defined merge tools" or
>> similar?
>>      
>
> IMHO yes.  After experimenting with transplant last week, I realized
> that Mercurial has first-class merges and second-class merges.  "hg
> merge" is first-class because it invokes your chosen merge tool(s): no
> .rej files to worry about.  "hg transplant" is second-class because
> conflicts result in .rej files.  Yuck.  As a result, cherrypicking
> fixes with Mercurial can actually be *worse* than with CVS, where at
> least you get inline merge markers (IMHO easier to work with than .rej
> files).
>
> "hg export | hg import -" is also second-class.
>    

and "hg qpush"

(And related to (my interpretation of) recent discussions about diff / 
extdiff: Diff can be considered a special case of merge with one parent. 
(AFAIK that is how the Mercurial data model sees it.) Often the same 
tools can be used for diff and merge. And it would make a lot of sense 
if diff in a working directory with two parents could invoke a "merge 
tool" with 3 panes: the two parents and the working directory. (Sune 
Foldager posted some extdiff patches related to that some months ago.) 
So: I think diff/extdiff should be added to the list of second-class 
citizens as well.)

/Mads

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://selenic.com/pipermail/mercurial/attachments/20090327/272038fe/attachment.htm