Re: [Monotone-devel] [PATCH] Make --execute default

[Markus Schiltknecht]

Hi,

William Uther wrote:
>   I've just finished a patch to make --execute the default.

Cool.

> Actually, 
> it entirely removes the --execute option, and adds a new option 
> --bookkeep-only which is approximately the inverse of -e.  This also 
> adds some checks to the drop implementation so that if you try to drop a 
> changed file (or a dir that isn't empty when it gets to it) then it 
> noisily falls back to --bookkeep-only.  This patch also changes rename 
> so that it warns when you're moving something and the source is missing 
> but the destination exists - it used to just silently fall back to 
> --bookkeep-only in that case.

Uhm... I recall talking about that on the summit. I didn't exactly 
follow the complete recent thread...

Did you consider all corner cases? Does monotone now act consistently 
for all commands? I.e.:

mtn drop ${STH}:
 - file ${STH} exists
 - dir ${STH} exists and is empty
 - dir ${STH} exists and is not empty
 - dir ${STH} does not exist

mtn add (--bookkeep-only)?  ${STH}:
 - file ${STH} exists
 - dir ${STH} exists
 - no 'something' exists - here, monotone currently bails out 
correctly. But do we want to support 'add --bookkeep-only'? How do we 
know if ${STH}  is going to be a directory or a file? (Probably we don't 
want to support that?)

mtn rename S{SOURCE} ${TARGET}:
 - uff... any combination of ${SOURCE} exists or not and ${TARGET} 
exists or not.

Please elaborate on those corner cases.

Do we want to call that argument '--bookkeep-only'? That seems ugly to 
me. And I don't recall naming what monotone does 'bookkeeping', anywhere 
else.

>   I've also fixed all the tests to make this work.  ugh.  They all pass 
> now (except for ws_ops_with_wrong_node_type which didn't pass before and 
> probably shouldn't be passing.  I think changing monotone so it passed 
> would be a mistake - the broken part of the test should be removed).  In 
> most cases I did the simple fix of just removing "-e" or adding 
> "--bookkeep-only", even when there was a move with "--bookkeep-only" 
> followed immediately by a rename.  I wanted to be conservative with the 
> test suite.

That sounds good.

>   The test restrictions_with_renames_and_adds was also failing before 
> this patch.  I fixed it so it passes.
>
>   I tried sending the patch to the list, but it was too large (the meat 
> of the patch is fairly small, but there were a lot of tests to fix up).  
> I've put it on the web (with my other monotone patches) here:

Why don't you simply get a monotone key? Send the public key to 
Nathaniel, he will happily give you push access to venge.net...

Regards

Markus