On Mon, Aug 3, 2009 at 07:07, Nick Sabalausky<business4_at_semitwist.com> wrote:
> levyam Full name Andy Levy Date 2009-08-03 03:29:35 PDT Message On Sun, Aug
> 2, 2009
>>
> "Corrupted" and "data is lost" implies that you took all the correct
> actions, but the software scrambled/lost bits. What you're describing
> is a function of user action which can easily be avoided by correct
> user behavior. As I said previously, you're given plenty of
> opportunities to review what you're committing before you do it.
> <
>
> "avoided by correct user behavior"? That is an absolutely preposterous,
> barely-above-trolling argument on the same arrogance level as "RTFM". The
> whole point of version control (as well as innumerable computer
> technologies) is to prevent various forms of user error! To presume that
> preventing user error is something that should not be done is completely
> ignorant. In fact, why are you even paying *any* attention to version
> control systems if you truly feel that way about user error? Let the actual
> authors of this project respond for themselves instead of continuing in your
> attempts to troll this legitimate request into oblivion.
Subversion does exactly what you tell it to. If you issue a command
that results in your commit not including what you think it should
include, then you issued the wrong commit. And you can easily roll it
back.
No version control system can protect the user from himself. What you
consider "corruption" and "data loss", Subversion considers a
consistent repository state because because the commit transaction
didn't fail, all checksums were verified, etc. Subversion doesn't
understand context, only bits.
>>
> Let's suppose for a moment that this option does exist, and in the
> course of your "juggling a million things" you commit your whole WC,
> committing changes that you didn't want because they were in the WC
> and you didn't notice, just like you've been "corrupting" your data
> today because you're committing too deep in the WC and not getting
> changes higher up in the tree? What will you do then?
> <
>
> I have never had a working copy that included both changed files I was ready
> to commit and changed files I wasn't ready to commit. In fact, I don't work
> in such a way that that would ever happen.
And I do all the time, because my change management process & timing
requirements more or less require that I do so (otherwise I'd need to
maintain a half-dozen WCs of a single project on my workstation). And
I've never had any problems with Subversion or TSVN resulting from it.
Every one of my commits has had exactly what I wanted and expected
included in it, because I review what's being committed before I click
OK.
------------------------------------------------------
http://tortoisesvn.tigris.org/ds/viewMessage.do?dsForumId=4061&dsMessageId=2379514
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_tortoisesvn.tigris.org].
Received on 2009-08-03 14:26:36 CEST