On 1/28/2007 4:06 PM, Ryan Schmidt wrote:
> On Jan 28, 2007, at 14:58, Eric wrote:
>
>> At 06:43 AM 1/28/2007, Duncan Murdoch wrote:
>>
>>> I'd disagree with that. One of the strong points of Subversion is
>>> that it does atomic operations. One of the things I dislike about
>>> Subversion is that only some of the operations are atomic. For
>>> example, it's easy to get in a mess when you do an update and it
>>> fails half-way through, leaving you with a mixed revision working
>>> copy.
>> Now, THIS I don't understand. Operations are either atomic, or
>> they are not. Everything I've read about Subversions makes a VERY
>> big deal out of the fact that operations are atomic. Now, are you
>> saying that they are only atomic "sometimes"?
>
> Some operations, like commits, are atomic. If any part of the commit
> fails, the entire commit fails. This is good.
>
> Other operations, like updates, are not atomic. If the update fails
> part of the way through -- for example because of temporary network
> problems -- part of the working copy is updated, and part of it is
> not. This is not ideal, but in many circumstances is not a big deal:
> just run the update again.
For me the most common cause for an update failure is that a file has
been added to the repository but an unversioned copy of the same file
exists on my disk. I need to rename/remove it before I can complete an
update, and then svn tells me about the next collision. Grrrr.
Duncan Murdoch
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Mon Jan 29 00:35:52 2007