[sql] Complain about statement mutations after stepping has started.

[sql] Complain about statement mutations after stepping has started.

DCHECK if any Bind*() calls occur after Step() has been called.  The
bind cannot be implemented at that point, so most likely the calling
code has an error, such as a forgotted Reset().

Also DCHECK if Run() is called after Run() or Step() without an
intervening Reset().  Such a calling sequence is unlikely to be
intentional, because the results of calling Run() twice aren't
defined.

BUG=309759
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=231098

[Scott Hess - ex-Googler, 2013-10-24 19:51:36]

Earlier, I was thinking "Hey, this is sorta gross, maybe it could be dressed up
with a helper or something."

Now I'm thinking "Whatever, it's like a 300-line file of glue code, helper
factory services aren't going to turn it into anything amazing."

https://codereview.chromium.org/40733003/diff/1/sql/statement.cc
File sql/statement.cc (right):

https://codereview.chromium.org/40733003/diff/1/sql/statement.cc#newcode88
sql/statement.cc:88: bool Statement::Succeeded() const {
Almost tempted to put:
  DCHECK(stepped_);
here.

[Greg Billock, 2013-10-24 21:49:50]

lgtm

https://codereview.chromium.org/40733003/diff/1/sql/statement.cc
File sql/statement.cc (right):

https://codereview.chromium.org/40733003/diff/1/sql/statement.cc#newcode88
sql/statement.cc:88: bool Statement::Succeeded() const {
On 2013/10/24 19:51:37, shess wrote:
> Almost tempted to put:
>   DCHECK(stepped_);
> here.

Hmmmm. I hear you. Maybe a follow-up try run with that set would shake loose a
bug or two...

[Scott Hess - ex-Googler, 2013-10-25 17:50:36]

thanks!

https://codereview.chromium.org/40733003/diff/1/sql/statement.cc
File sql/statement.cc (right):

https://codereview.chromium.org/40733003/diff/1/sql/statement.cc#newcode88
sql/statement.cc:88: bool Statement::Succeeded() const {
On 2013/10/24 21:49:50, Greg Billock wrote:
> On 2013/10/24 19:51:37, shess wrote:
> > Almost tempted to put:
> >   DCHECK(stepped_);
> > here.
> 
> Hmmmm. I hear you. Maybe a follow-up try run with that set would shake loose a
> bug or two...

Fails SQLStatementTest.Run :-).

Thinking more about it, I'm no longer convinced about this notion.  Succeeded()
is documented as working this way.  I could kind of see a case where your
control-flow is such that it has a few cases, some of which optionally fire a
Step(), and then at the end it asks whether the statement has made any progress.

[I think that the is_valid() test may not be necessary, here.  succeeded_ can
never get to true if the statement isn't valid.]

[commit-bot: I haz the power, 2013-10-25 17:51:55]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/shess@chromium.org/40733003/120001

[commit-bot: I haz the power, 2013-10-25 20:09:33]

Change committed as 231098