Re: PATCH: Precise control over query's modification state

Hi

On Wed, May 21, 2014 at 6:44 PM, J.F. Oster <jinfroster(at)mail(dot)ru> wrote:
> Hello,
>
> Regarding this report:
> http://www.postgresql.org/message-id/C61C1DE0-A5ED-4318-9391-7E35081C56D5@empiric.al
>
> AZ>> 7. When I open a SQL console while an object/function/etc is
> AZ>> selected, the text inside is how to create it. That’s
> AZ>> (occasionally) helpful. However, if I close it without modifying
> AZ>> it, I shouldn’t be asked if I’m sure I don’t want to save it first
> AZ>> when I haven’t edited anything inside yet.
>
> JFO> That's the way an option of query editor "Copy SQL from main window to
> JFO> query tool" is implemented. It makes an initial modification of empty
> JFO> textarea for you, surely marking it as modified.
>
>
> Attached patch adds member variable to track last origin of query
> modification:
> ORIGIN_MANUAL - edited directly by user
> ORIGIN_FILE - loaded from file
> ORIGIN_INITIAL - passed into constructor
> ORIGIN_HISTORY - picked from "previous queries"
> ORIGIN_GQB - generated by query builder
>
> Such approach allows for some enhancements of user's experience.
> I've implemented the following in this patch:
>
> 1. Text initialization in constructor isn't considered a change, so it
> will not ask for save confirmation if left intact.(*)
>
> 2. Allow to "Save As" the initial text immediately, without having to
> modify text in order to enable save button. Also the behaviour
> shouldn't depend on settings->GetStickySql(), because there are other
> use-cases not related to this option (for example, insert-scripts on a
> table).
>
> 3. Skip confirmation to replace query text by GQB if previously it was
> generated by GQB and was left intact.

Thanks - patch applied (seems fine on Mac too).

> (*) - I've run into somewhat inconsistent behaviour of STC under
> Windows - some more events may come after query initialization and
> wxSafeYield() in constructor. These late events switch origin from
> initial to manual, thus breaking the idea of #1. I've made a hack to
> filter these events out and it seems to work ok (frmQuery.cpp:1692) on
> Windows and Linux.
> But think I'm missing the point in working with STC. Probably it's a
> separate thread. I don't know how to work around that better :(

STC is a little weird because it's a third party project that's
included with a wrapper in wxWidgets. Sometimes, we have to just live
with its little quirks :-/

--
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake

EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company