REBOL/Command vs Doc's MySQL scheme
[1/8] from: mat:plothatching at: 31-Mar-2003 8:57
Hello Gregg,
GI> Shouldn't be too different, at least to see which works best for you,
GI> and as Tim said, it works with Core.
Inside of a few minutes, I'd turned my script into using DocKimbles's
MySQL scheme. The results were... surprising.
Take a look at these timings;
REBOL/Command;
>> system/version
== 2.0.0.3.1
>> do %listsql.r
Script: "Untitled" (none)
SQL database openned.
>> databaseinit
Initialised vw1...
Initialised vw2...
Initialised vw3...
Initialised vw4...
0:00:01.742
>> importallcsv
0:01:28.157
Nearly 2 seconds to create four tables. A minute and a half to import
400 odd records from a CSV.
Now let's look at REBOL/Core with DocKimbles MySQL scheme;
>> system/version
== 2.5.5.3.1
>> do %listsql2.r
Script: "Untitled" (none)
Script: "mySQL Protocol" (25-Jul-2001)
mySQL protocol loaded
connecting to: wench
SQL database openned.
>> databaseinit
Initialised vw1...
Initialised vw2...
Initialised vw3...
Initialised vw4...
0:00:00.03
>> importcsvall
** Script Error: importcsvall has no value
** Near: importcsvall
>> importallcsv
0:00:00.751
Errr.... That's 117 times faster than REBOL/Command.
*Madness* Still, problem solved! :)
Regards,
Mat Bettinson - +44-(0)20-83401514.
[2/8] from: antonr:iinet:au at: 31-Mar-2003 19:23
Yes, I think I remember some people
mentioning that Rebol/Command has gotten
a bit out of date and was buggy ?
But the suggestion was to contact Rebol Tech
for a newer beta. ???
Ask the guys on AltME in the REBOL/Command world.
Anton.
[3/8] from: petr:krenzelok:trz:cz at: 31-Mar-2003 11:29
Mat Bettinson wrote:
>Hello Gregg,
>GI> Shouldn't be too different, at least to see which works best for you,
<<quoted lines omitted: 57>>
>0:00:00.751
>Errr.... That's 117 times faster than REBOL/Command.
Surprising, isn't it? I remember Maarten or someone else stating similar
measures some time ago .... I wonder what technique does Command's MySQL
uses. But - except for security and maybe ODBC, there is not much value
in investing into Command nowadays .... Our company did so, because we
need ODBC and it works very well for us, but FastCGI (under Windows) is
half-functional and mySQL performance reported as being worse than even
Doc's script-only solution. What is more, Doc adds PostGress driver and
if he would not want to "hurt" RT, we would have even proper FastCGI stack.
Maybe SoftInnov should be contracted to program/strenghten some
Command's "connectors" :-)
-pekr-
[4/8] from: Maarten:Koopmans:surfnet:nl at: 31-Mar-2003 13:50
Matt,
I reported this a year ago or so. Doc's scheme is very good indeed. The
difference with the command implementation is... painful.
At the time we started using MySQL (with InnoDB transactional backend)
in stead of Oracle, and then turned to Doc's driver. This improved "out
of the box" performance twice with a factor 30. Of course you can tune
Oracle, but we were in the proof-of-concept business.
Now I know our measurements were correct.
--Maarten
Petr Krenzelok wrote:
[5/8] from: greggirwin:mindspring at: 31-Mar-2003 8:55
Hi Mat,
Glad to hear there was an easy solution. I think Pekr has said before
that Doc's is a much better way to go and that RT should ship his
instead. :)
-- Gregg
[6/8] from: mat:plothatching at: 31-Mar-2003 18:49
Hello Gregg,
GI> Glad to hear there was an easy solution. I think Pekr has said before
GI> that Doc's is a much better way to go and that RT should ship his
GI> instead. :)
Indeed, I mean my application really was nothing requiring speed but
117 times slower meant it was holding even that back!
Still, I'm very happy now. In a few days I've gone from not knowing
anything about SQL and having never written a proper database
application to having one up and running that I'm very pleased with.
I had to write some other things for this while I was at it, including
a proper dual plain/HTML mail sender with in-lined embedded image
attachments (yes yes I know but the client likes this stuff). That
might actually be quite a good one to supercede the one that's in the
library at the moment.
Regards,
Mat Bettinson - +44-(0)20-83401514.
[7/8] from: tim:johnsons-web at: 31-Mar-2003 10:46
* Mat Bettinson <[mat--plothatching--com]> [030331 10:02]:
> Hello Gregg,
>
> GI> Glad to hear there was an easy solution. I think Pekr has said before
> GI> that Doc's is a much better way to go and that RT should ship his
> GI> instead. :)
Hello All:
---------
Operative word 'ship:
I can only speak for the components that I have used and
that is Andrew's ML dialect and Doc's MysqlProtocol.
(my apologies to others whose components I have not used)
Yes, perhaps RT should consider 'shipping' user built
components that are industrial strength.
I am voting for Mysql Protocol and the ML dialect. This
is from my own experience...
MTCW: tim
> Indeed, I mean my application really was nothing requiring speed but
> 117 times slower meant it was holding even that back!
<<quoted lines omitted: 12>>
> [rebol-request--rebol--com] with "unsubscribe" in the
> subject, without the quotes.
--
Tim Johnson <[tim--johnsons-web--com]>
http://www.alaska-internet-solutions.com
http://www.johnsons-web.com
[8/8] from: petr:krenzelok:trz:cz at: 31-Mar-2003 20:45
Gregg Irwin wrote:
>Hi Mat,
>
>Glad to hear there was an easy solution. I think Pekr has said before
>that Doc's is a much better way to go and that RT should ship his
>instead. :)
>
>-- Gregg
>
Hey, wait a bit ;-) It was NOT me who claimed RT's mySQL protocol is
underpowered :-) I even have not tried mySQL with Command yet, only ODBC
... so ... :-)
Cheers,
-pekr-
Notes
- Quoted lines have been omitted from some messages.
View the message alone to see the lines that have been omitted