Piotr Przybylski a écrit :
2011/8/4 Ammar Yasir ayax88@gmail.com:
On Thu, Aug 4, 2011 at 2:01 AM, Marc Delisle marc@infomarc.info wrote:
Ammar Yasir a écrit :
On Wed, Jul 27, 2011 at 10:22 PM, Ammar Yasir <ammaryasir.88@gmail.com mailto:ammaryasir.88@gmail.com> wrote:
On Wed, Jul 27, 2011 at 3:37 PM, Marc Delisle <marc@infomarc.info <mailto:marc@infomarc.info>> wrote: Ammar, With Firebug I had a look at the network traffic when I click a
data point to edit it: I was surprised to see none.
IMO this is not good: it means that all the columns for all rows are in memory, making the browser able to handle far less rows. Is there a reason why you are reading the complete rows to generate the plot? I expected that you would just read the necessary columns, then use AJAX to read a complete row when the user wants to edit it. I'm currently working on the edit feature for strings. I'll work on it after this. --
Implemented this and pushed to my repo. I only send the xField, yField and dataLabel to the user now.
Ammar, you have reduced the amount of data transferred between the PHP level and the Javascript level, by sending less data in the querydata div.
Can you also reduce what is transferred between the MySQL server and the web server, by avoiding to generate "SELECT *" in the query generation part, instead selecting only the needed columns?
That would have been the ideal case but to generate the unique-condition( function PMA_getUniqueCondition($handle, $fields_cnt, $fields_meta, $row, $force_unique=false) ), I need the complete row. So I cannot do much about the query generation on the server side other than putting a limit on the number of rows retrieved.
It will increase complexity o bit but you can do one of the two:
- select one full row with "SELECT * FROM ... LIMIT 1" (any full row
will do so no ORDER here) and analyze which rows you need by looking for primary/unique key fields,
- if you have only one table you can analyze SHOW CREATE TABLE
statement and look for PRIMARY KEY and UNIQUE constraints
But PMA_getUniqueCondition() also handles the case when there is no unique key.
That way you can do the same filtering that PMA_getUniqueCondition does when building selection condition. If you find primary key or unique key you can use only these columns, otherwise SELECT * is the only way.