[Phpmyadmin-devel] Inline edit and UPDATE feedback

[Marc Delisle]

Le 2011-03-20 07:26, rohit sharma a écrit :
> On Sun, Mar 20, 2011 at 2:57 PM, Marc Delisle <marc at infomarc.info> wrote:
> 
>> Le 2011-03-19 17:26, rohit sharma a écrit :
>>> On Sun, Mar 20, 2011 at 2:36 AM, Marc Delisle <marc at infomarc.info>
>> wrote:
>>>
>>>> Le 2011-03-19 17:03, rohit sharma a écrit :
>>>>> On Sun, Mar 20, 2011 at 2:15 AM, Marc Delisle <marc at infomarc.info>
>>>> wrote:
>>>>>
>>>>>> Le 2011-03-19 16:42, rohit sharma a écrit :
>>>>>>> On Sun, Mar 20, 2011 at 2:03 AM, Marc Delisle <marc at infomarc.info>
>>>>>> wrote:
>>>>>>>
>>>>>>>> Le 2011-03-19 16:21, rohit sharma a écrit :
>>>>>>>>> On Sun, Mar 20, 2011 at 1:45 AM, Marc Delisle <marc at infomarc.info>
>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>>> Hi,
>>>>>>>>>> by default, the ShowSQL directive is true, which makes the
>> generated
>>>>>> SQL
>>>>>>>>>> appear on top for almost every action. This fulfills many needs:
>>>>>>>>>> - people teaching SQL
>>>>>>>>>> - user who wants to verify what phpMyAdmin did in the backend
>>>>>>>>>>
>>>>>>>>>> However with Inline edit, we no longer see this generated SQL and
>>>> I'm
>>>>>>>>>> wondering if something should be done about this.
>>>>>>>>
>>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> We can add a field for the last inline_edit query run by the user,
>>>> and
>>>>>>>>> display it using a show/hide button. This field can be added just
>>>>>> beneath
>>>>>>>>> the "generatedSQL" which is always displayed.
>>>>>>>>
>>>>>>>> Hi,
>>>>>>>> in the case of Edit, there is no Show/hide because the interface
>>>>>>>> respects the $cfg['ShowSQL'] directive, so why should there be a
>>>>>>>> Show/hide in the case of Inline Edit?
>>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> In case of Edit, the query tends to be small, "select * from
>>>> table_name;
>>>>>> "
>>>>>>> (talking generally, although the query can be much longer) , and on
>> the
>>>>>>> other hand inline-edit queries are update queries which can be quite
>>>> long
>>>>>>> for tables with more columns, hence it would take away some valuable
>>>>>> space
>>>>>>> from the displayed table.
>>>>>>
>>>>>> I'm not sure I follow you. If you browse a table, click Edit on some
>> row
>>>>>> and make a change, you'll see the generated UPDATE statement, followed
>>>>>> by a SELECT representing your current results set.
>>>>>
>>>>> Hi,
>>>>>
>>>>> Thanks. I got the idea. But in the case of "inline_edit" the query is
>>>> being
>>>>> sent using Ajax, and the frame won't  move after a query execution, so
>>>> where
>>>>> do your propose to display the inline_edit query that was last run, so
>>>> that
>>>>> the user gets notified.
>>>>
>>>> At the same place where it's displayed when using the non-Ajax Edit (but
>>>> obviously not with the same mechanism).
>>>>
>>>> But my initial question was not about implementation, it was about
>>>> whether we should display UPDATE feedback at all for Inline edit.
>>>>
>>>> --
>>>> Marc Delisle
>>>> http://infomarc.info
>>>>
>>>
>>> Hi,
>>>
>>> After inline_edit, the user instantly gets notified of the change in the
>>> table since the position of the frame remains unchanged. Although it can
>> be
>>> a good idea to display the query for users who wish to know what all
>>> happened in the backend and also, if my query is incorrect, I would like
>> the
>>> error and the query to be displayed for a longer time (permanently)
>>  rather
>>> than for some short duration.
>>
>> Thanks for your feedback. I believe that the UPDATE query should be
>> displayed, if it works or not, at the same place than when using a
>> traditional edit (if $cfg['ShowSQL'] is true).
>>
>>>
>>> Also, if my inline_edit query is incorrect, a ajax window displays the
>>> error, but the "back" button on that window takes me all the way to
>> "insert
>>> new record" appending the query in my URL, shouldn't it direct me to the
>>> same page as before ?
>>
>> Look on Google for "Ajax back button", you'll see that in general, back
>> button with Ajax is problematic.
>>
> Hi,
> 
> Then  the  "back" button makes no sense, because the user would like to stay
> on the same page and look at what went wrong, rather than going to any other
> page.

I don't understand your point. Why would the user click Back in this
case? He is on the same page and should be able to fix the problem, or
click Hide to quit inline editing.
> 
> 
>> I think that we should concentrate on not generating a bad query for
>> inline edit. Do you have a test case where inline edit generates wrong SQL?
>
> I think when dealing with foreign keys one can come up with incorrect update
> statements. In that case, the error message being displayed using
> PMA_ajaxShowMessage() isn't helpful enough as it's just transient.

Would it be better to show it during 10 seconds?

-- 
Marc Delisle
http://infomarc.info