World: r3wp

I think I mean 
   where prod-code in
not NOT in
(the code is not tested, and may not be possible in mySQL)

Joe Celko's SQL for Smarties is probably the best book for getting 
up to speed.

dunno if newer mysql supports subselects, but it didn't in the past

4.1 seems to support them. Dunno if they work though :-)
http://dev.mysql.com/doc/mysql/en/subqueries.html

I need to learn SQL more deeply

 A broad introduction can be found here: http://riki-lb1.vet.ohio-state.edu/mqlin/computec/tutorials/SQLTutorial.htm

I'm sure one of you guys knows how to check so that I can avoid:
** Access Error: Port rhm not open
** Where: read-rows
** Near: a: copy db

I think this happens after I have inserted a query and there are 
no results, but in any case I'd like to able to check a port to see 
it is open.

I should add that the code looks like this ...and it works when a 
record is found:
    insert db reduce [s]
	a: copy db

while copy db

foreach record copy db [ probe record ]

Graham, thanks. So this would check if there was something returned 
by the query. I'll check it out.

I have a problem when the connection times out and mysql-protocol.r 
wants to reconnect. It looks like it hangs, but on closer inspection 
with trace/net on just gives an infinite amount of low-level reads 
of length 0 bytes.

I can close and reopen the connection, and it works again, but I 
can only do this 5-6 times before it refuses to connect to the mysql 
server. Is this a known problem?

You can try to disengage the auto-reconnect feature and test if the 
connection is still active by yourself (send a ping command for example).

db-port/locals/auto-ping?: off (stop pinging the server and don't 
reconnect automatically).

Henrik: I hit the same problem just at this moment:( unfortunatelly 
I don't know how to fix it

But I'm not sure if it's not problem of the MySQL itself (for example 
some way how to awoid attacks

I found a different solution: manually check if the connection has 
timed out 30 seconds after a query and then reconnect. it's not perfect 
since I have to check at every query, but it works to the extent, 
I've tested it

how you do the timeout test?

after the query I store 'now. the next time I do a new query I check 
'now against the stored one. if the difference is more than 30 seconds, 
I close the db connection and open it again immediately

that would not solve my problem

my problem was that after about 30 queries it got to infinitive loop 
(low-level reads of length 0 bytes)

What's your server version ?

MySQL server version: 3.23.37

an atomic way of returning the value of an AUTO_INCREMENT column 
on insert  would be very nice.

i.e. 

insert db "insert  into tab(tab_id,tab_name) values(NULL,'alt_key')"
tab-id: copy db


would gaurentee to give you the correct tab_id   as it looks now 
you have to:


insert db "insert  into tab(tab_id,tab_name) values(NULL,'alt_key')"
;;; hope no one else is insertintginto the table right now
insert db  "select LAST_INSERT_ID()"
tab-id: copy db

even being able to issue more than one sql statement would help some

insert db {insert  into tab(tab_id,tab_name) values(NULL,'alt_key'); 
select LAST_INSERT_ID()}
tab-id: copy db

Tom: the select above is actually safe. MySQL keeps the last insert 
id on a per-connection basis

this means, that even if someone else inserts a record between your 
two queries, you'll get the correct result.

each connection gets its own last insert id.

thanks Gab, that makes me feel better

Hello, is there anyone working with Rebol Command and Mysql 4.1.x 
? It appears that the authentication protocol has changed and Rebol/Command 
can not connect to MySQL 4.1.x. But it works fine with MySql 4.0.x.

have you tried free mySQL driver from DocKimbel?

I have heard that the driver from DocKimbel has the seem issue... 
But my point is that Rebol/Command should be updated to work with 
version 4.1 which is currently the recommended version for peoduction 
environment!

Most probably driver support for db will be drop in future versions. 
(I recommend working with DocKimbel of helping improve his code).

(Disclaimer: This is my opiong not RT official position)

If support for db is gone, what is the point for a professional version 
of rebol??

It is going to become platform. You get access to a plugin API where 
you can plug-in 3rd db modules. (At leass the was the plan from DevCon'04). 
Don't know how things will change in the future. You can still request 
RT to upgrade the drivers.

What exactly is Platform going to be about?

Imo we should aim for REAL kernel - where all unnecessary things 
should be removed - I talk about fastcgi, db drivers etc.

I am not sure about protocol handlers either ... I start to not understand, 
what is rebol - it is not language, nor it is stand-alon platform 
... I want clean design ... Base + SDK is the answer imo ...

Base == Platform imho

good then ...

so we should be able "to build" various variants upon our (developers) 
needs using SDK modules ...

Ok, but who will develop those modules? Would they be open source? 
Personally, the reason why I love rebol is because it allows me to 
focus on the business, not on the tehnical aspect of a database, 
fastcgi, etc... I am certainly not the one who will develop technical 
modules, simply by lack of knowledge and time. But I really need 
those db access, fastcgi, etc...

It could be RT, or they could come from 3rd parties.  I guess the 
answer is time will tell. You can certainly express your views and 
requests to RT, they may change direction based on customer input. 
However the directions is a bit blurry right now.

Is there any conclusion about DocKimbel's mySQL driver and 4.11 server 
release? Does it work or no? In terms of one project, where we exchange 
data, client agrees to our proposed way, only if we import data to 
their tables on their server. It is Debian linux, mySQL 4.11 or so, 
so before I produce some script and give it to them, I would like 
to ask if the problem of connection/authentication is resolved? Thanks 
a lot ....

Can I update particular fields (columns) in table?

Hmm, so noone uses mySQL here? :-)

I would have to try to download it, install and try it myself, uhm, 
another lost afternoon for testing simple thing :-)

the trouble is, I need to send rebol plus script to other site, so 
testing will be very tough .... hmm, maybe there is no other better 
way than to actually do so ....

btw - other site will use rebol plus mysql script in terms of Cron 
job - how should I launch it? I hope they know how to define Cron 
job, as I don't know ... should it be similar to CGI script? Or? 
Simply "cron script" calls something like \path\to\rebol\rebol \path\to\my-script\myscript.r? 
What arguments eventually? -s?