Patch for issue #1058

[cstiborg]

This small patch assures that the indicators_ vector is cleared after its content is deleted in bind_cleanup(). If this is not done the indicators vector grows indefinitely as it caches the deleted and NULLed indicators. See issue #1058 for more explanation/comments.

[vadz]

Thanks for the PR, I agree that it looks like the right thing to do, i.e. I don't see why should we have an ever-growing indicators_ vector, but I don't understand how can just clearing it, without adjusting any indices for accessing it anywhere, not break something: it looks like either the indices into it were wrong before (always a possibility, of course) or they must be wrong now.

But OTOH I don't actually understand where is this indicators_ actually used at all. It is filled in statement_impl::bind(values & values), but where is it used then? I wonder if we could just remove this completely, i.e. could it be a leftover from some earlier version? Or am I missing something obvious here?

[zann1x]

Just my two cents:

But OTOH I don't actually understand where is this indicators_ actually used at all. It is filled in statement_impl::bind(values & values), but where is it used then? I wonder if we could just remove this completely, i.e. could it be a leftover from some earlier version? Or am I missing something obvious here?

That's what I wondered about as well. But I think that statement_impl::indicators_ serves as kind of cleanup bucket for values::indicators_ because the entries don't seem to be cleared anywhere else. But the reason for this indirection is unclear to me as well.

[vadz]

That's what I wondered about as well. But I think that statement_impl::indicators_ serves as kind of cleanup bucket for values::indicators_ because the entries don't seem to be cleared anywhere else.

Ah, indeed, thanks, you're absolutely right.

But the reason for this indirection is unclear to me as well.

I have no idea what is the life cycle of values supposed to be, so I guess it might be too short?

Anyhow, I'll merge this PR for now, thanks again to both of you!

[vadz]

Merged in the commit above.