MQ usability

[Peter Williams]

On 16/08/09 20:08, Dirkjan Ochtman wrote:
> On Tue, Aug 11, 2009 at 02:18, Peter Williams<pwil3058 at bigpond.net.au>  wrote:
>> qsave and its usefulness in updating ones patches when the underlying code
>> changes (see page 197 of "Mercurial, The Definitive Guide",
>> ISBN-978-0-596-80067-3) is the MQ feature that most distinguishes from other
>> patch management systems (such as the very popular "quilt"
>> <http://savannah.nongnu.org/projects/quilt>).  So it would be a big mistake
>> to remove it.
>
> You could just use the rebase extension to rebase applied patches to
> the new tip. Seems much easier to use and remember (and is more
> accessible, IMO).

I was unaware of the existence of 'rebase' but now that I am I'll give 
it a try.  I'll admit that the procedure described for 'qsave' et al is 
a little complicated which is why I put the special menu (for it) in 
gwsmhg's MQ page.

However, I think that if it does do the job then a command 'qrebase' 
based on it (but tailored to MQ's specific needs) be incorporated into 
MQ before 'qsave' is removed.  (I imagine that its interface would be 
simpler than rebase's as there should be less things to specify.)  MQ 
needs to have STAND ALONE functionality that is (at least) the 
equivalent of its current functionality.

>
>> Then what would be the point of having the<file>  argument if qrefresh
>> behaves the same with it as without it.
>
> It doesn't: without<file>  it will save changes for all files, with
> <file>  it will only save changes to that file to the patch and leave
> other files modified in the working dir.

Yes, but the proposal (that I was responding to) was to change that (as 
I understood it) so that there was no difference.  Hence my question. 
If I misunderstood the proposal, I apologise.

Peter
-- 
Peter Williams                                   pwil3058 at bigpond.net.au

"Learning, n. The kind of ignorance distinguishing the studious."
  -- Ambrose Bierce