Dave Page a écrit :
> On Sat, Jun 14, 2008 at 3:01 PM, Guillaume Lelarge
> <guillaume@lelarge.info> wrote:
>
>> Finally, here is the prototype. As I first talked about this one year ago, I
>> will summarize the idea : adding a checkbox on the SQL tab of an object's
>> properties to let the user change the SQL query. Checking will disable the
>> contents of the other tabs because we don't want to try to reverse engineer
>> the user's changes.
>>
>> So, here is the patch that does this. I'm sure there's work left to do (most
>> notably some duplicate code) but, at least, it works for me on two different
>> scenarios : changing the SQL query and adding another SQL query.
>
> I haven't tested yet (just reviewed the diff), but a couple of
> thoughts come to mind:
>
> - Can we disable the controls on the form, rather than the tabs so the
> user can still browse the object details after modfying the SQL? (Or
> does disabling the tab effectively do that?). The downside of that is
> that we'd need to keep track of which controls were already disabled
> when we disable them all, so if re-enabling them we end up in the
> original state. I realise this is not what I suggested previously.
>
You can't really disable a tab. You disable the page of the tab, so you
can still browse the different tabs.
> - Before returning to GUI mode, we should warn the user (if he has
> modified the SQL) that his changes will be lost. Of he accepts the
> change, the SQL should be regenerated immediately.
>
The patch does not warn the user, but I think it would be a good
addition. But the SQL is regenerated as soon as the user clicks on the
checkbox a second time.
> So, before I test this on my lapdog, I'm sure we all want to know -
> did anything actually explode?
>
AFAIK, no. Nothing exploded.
--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com