[sqlite] How to use SQLite with Bold (Delphi Architect)

[[EMAIL PROTECTED]]

I put it all on a web-page, also the needed binaries.

http://www.rosa.nl/Bold2/

Kind regards
Bert Verhees

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Darren Duncan]

At 11:49 AM -0500 7/23/04, Greg Miller wrote:
There are certainly a few drawbacks to using Reply-To, but I can't 
remember the last time I was on a mailing list that didn't make use 
of it. Clearly, there's a very solid consensus out there that 
Reply-To is the way to go.
That may be because a large fraction of email lists use something 
like Yahoogroups, which often uses reply-to-group as a default 
setting, though thankfully Yahoo lets you turn that option off. -- 
Darren Duncan

Re: [sqlite] 3.x pragma synchronous

[Guillaume Fougnies]

Yes, i'm fine with the synchronous=YES -> synchronous=FULL.
It's the good choice.
But in my case, i always need synchronous=OFF.
I'm working on an high availability project using the model
of Extended Virtual Synchrony. For the speed of
the distributed system during update/insert, i need
to reduce in all nodes the group transaction time.
If a node failure happens, the recovery procedure will
trash and/or resync databases from valid nodes.
That's why i do not care of local disk sync.
For now, it's running with 2.8 because i was unable to
get the same speed with 3.0 and synchronous off... during
my tests, consequences are lots of busy timeout despite the
same busy handler.
I'm sad because concurrency and btree+ storage in 3.0 seems
nicely improved, a great work as far as i can see.


Fri, Jul 23, 2004 at 02:18:14PM -0400: D. Richard Hipp wrote:
> Guillaume Fougnies wrote:
> >
> > PRAGMA default_synchronous disappeared from 3.x. Should we now 
> execute a
> > PRAGMA synchronous each time a db is opened?
> >
> 
> Yes.  The thinking was that there is rarely ever a good
> reason to make a persistent database less than fully synchronous.
> So synchronous=FULL is now the default.  If you want something
> other than synchronous=FULL, you have to define it anew at the
> beginning of each session.
> 
> So tell me: Why do you not want synchronous=FULL?

--
Guillaume FOUGNIES

Re: [sqlite] Alternating Output

[Randall Fox]

On Fri, 23 Jul 2004 18:50:49 +0200, you wrote:

>
>> How do you renumber an index anyway?
>
>Hmm ... given the tables
>   work (idx, sortcrit)
>   storeroom(rowcount integer)
>
>where 'idx' may be the index to renumber and 'sortcrit' a sorting criteria.  
>
>I would try to do something like the code below, but I just remembered that 
>sqlite doesn't support recursive triggers, therefore this won't work :(

So much for theory, I guess the only way is to write a custom
function..

On a side note, is there a repository of user written functions
somewhere that I can browse?


Thanks

Randall Fox

[sqlite] db recovery

[Ara.T.Howard]

somehow i've managed to corrupt my db - the error was
  ./src/pager.c:1179: syncJournal: Assertion `pPager->nRec*pgSz+hdrSz==jSz' failed.
running PRAGMA integrity_check yields something like
  carp:~/eg/ruby/rq/rq-0.1.2 > sqlite q 'PRAGMA integrity_check' | head
  *** in database main ***
  On page 43 cell 0: Key is out of order
  On tree page 41: initPage() returns error code 11
  rowid 1 missing from index jobs_jid_idx
  rowid 2 missing from index jobs_jid_idx
  rowid 3 missing from index jobs_jid_idx
  rowid 4 missing from index jobs_jid_idx
  rowid 5 missing from index jobs_jid_idx
  rowid 6 missing from index jobs_finished_idx
  rowid 6 missing from index jobs_jid_idx
is there a way to recover?
-a
--
===
| EMAIL   :: Ara [dot] T [dot] Howard [at] noaa [dot] gov
| PHONE   :: 303.497.6469
| A flower falls, even though we love it;
| and a weed grows, even though we do not love it. 
|   --Dogen
===

Re: [sqlite] PENDING & WRLOCK/RDLOCK

[D. Richard Hipp]

b.bum wrote:
Some more locking questions...
When sqlite3OsLock() takes out the PENDING lock, it switches between 
RDLOCK and WRLOCK depending on the locktype requested.

However, it doesn't appear that any subsequent code checks for RDLOCK 
vs.WRLOCK on the PENDING BYTE.  Actually, it appears that the only GETLK 
test is against the RESERVED byte.   Where is PENDING ever tested for on 
the actual file?


When a process tries for a SHARED lock, it first gets a a RDLOCK on
the PENDING_BYTE.  It releases that RDLOCK after the (separate) RDLOCK
is obtained on the appropriate byte(s) in the shared lock region.
When a process gets a PENDING lock, it gets a WRLOCK on the PENDING_BYTE
and holds it until the PENDING lock is released.
The WRLOCK on the PENDING_BYTE prevents new SHARED locks from being
acquired while the PENDING lock is held.  This is designed to prevent
(or at least limit) writer starvation.
--
D. Richard Hipp -- [EMAIL PROTECTED] -- 704.948.4565

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Mrs. Brisby]

On Fri, 2004-07-23 at 10:25, D. Richard Hipp wrote:
> Dan Keeley wrote:
> > Or, why dont we have a vote, and once it's decided, tuff luck thats how 
> > it stays? :)
> > 
> 
> Better:  Somebody please write me a simple, secure, mail handler to
> replace qmail/ezmlm that lets each user decide for themselves whether
> they want a Reply-To back to the mailing list or unmunged headers.
> 
> I'll be happy to supply volunteers with a detailed specification of
> what I am looking for in a mail system.

I can't. Your mailing list messages prevent me from replying to you. :)

ezmlm _will_ do this, btw. Simply create sublists; one with munging and
one without.

Re: [sqlite] 3.x pragma synchronous

[D. Richard Hipp]

Guillaume Fougnies wrote:
>
> PRAGMA default_synchronous disappeared from 3.x. Should we now execute a
> PRAGMA synchronous each time a db is opened?
>
Yes.  The thinking was that there is rarely ever a good
reason to make a persistent database less than fully synchronous.
So synchronous=FULL is now the default.  If you want something
other than synchronous=FULL, you have to define it anew at the
beginning of each session.
So tell me: Why do you not want synchronous=FULL?
--
D. Richard Hipp -- [EMAIL PROTECTED] -- 704.948.4565

[sqlite] PENDING & WRLOCK/RDLOCK

[b.bum]

Some more locking questions...
When sqlite3OsLock() takes out the PENDING lock, it switches between 
RDLOCK and WRLOCK depending on the locktype requested.

However, it doesn't appear that any subsequent code checks for RDLOCK 
vs.WRLOCK on the PENDING BYTE.  Actually, it appears that the only 
GETLK test is against the RESERVED byte.   Where is PENDING ever tested 
for on the actual file?

Re: [sqlite] Alternating Output

[Daniel Franke]

> How do you renumber an index anyway?

Hmm ... given the tables
work (idx, sortcrit)
storeroom(rowcount integer)

where 'idx' may be the index to renumber and 'sortcrit' a sorting criteria.  

I would try to do something like the code below, but I just remembered that 
sqlite doesn't support recursive triggers, therefore this won't work :(

-- 8< --
CREATE TRIGGER renumber_work_table_index AFTER UPDATE OF storeroom 
WHEN new.rowcount >= 0
BEGIN
UPDATE work 
SET idx   = (SELECT rowcount FROM storeroom) + 1,
color = idx % 2
WHERE sortcrit == (SELECT (min(sortcrit) WHERE idx == 0) AND idx == 0; 

/* stop if storeroom.rowcount equals the number of rows in work */
UPDATE storeroom 
SET rowcount = CASE 
WHEN (SELECT rowcount FROM storeroom) == SELECT count(*) THEN -1
ELSE (SELECT rowcount FROM storeroom) + 1 
END;
END;

/* fires the renumber_work_table_index-trigger */
CREATE TRIGGER catch_inserts AFTER INSERT ON work
BEGIN
UPDATE work SET idx = 0;
UPDATE storeroom SET rowcount = 0;  
END;

/* fires the renumber_work_table_index-trigger */
CREATE TRIGGER catch_deletes AFTER DELETE ON work
BEGIN
UPDATE work SET idx = 0;
UPDATE storeroom SET rowcount = 0;  
END;
-- 8< --

Cheers
Daniel

-- 
Dipl.-Math. (FH) Daniel Franke
Institut fuer Medizinische Biometrie und Statistik
Medizinische Universität zu Luebeck
Ratzeburger Allee 160, Haus 4
23538 Luebeck
Telefon: 0451-500-2786
Telefax: 0451-500-2999
[EMAIL PROTECTED]

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Greg Miller]

Michael Roth wrote:
Please switch off Reply-To again.
There are certainly a few drawbacks to using Reply-To, but I can't 
remember the last time I was on a mailing list that didn't make use of 
it. Clearly, there's a very solid consensus out there that Reply-To is 
the way to go.

--
http://www.velocityvector.com/ | "F--- 'em all. That's how I feel."
http://www.classic-games.com/  | -- Michael Moore on small business
http://www.indie-games.com/|

[sqlite] 3.x pragma synchronous

[Guillaume Fougnies]

Hello,

PRAGMA default_synchronous disappeared from 3.x.
Should we now execute a PRAGMA synchronous each time a db is
opened?
Could it be possible to get a configure flag or a define
in sqliteInt.h to set this by default?

Regards,
--
Guillaume FOUGNIES

[sqlite] compiled binary on cygwin always results in "SQL error: database is locked"

[Jeff Dever]

SQLite Version 3.0.3
 
When I compile my own sqlite3.exe binary and try to create a table, I
get the message "database is locked".  This happens when I use the
configure script from the sqlite tarball, or when compiling the objects
seperately and linking with the sqlite-source zip file.  The same
message occurs if I specify a file or :memory: for both the command line
shell and when linking with libsqlite3.
 
If I use the precompiled .exe, it works as expected.  If I compile on
Linux, it works as expected.  Any ideas why this would happen with
cygwin?
 
$ sqlite3 ":memory:"
SQLite version 3.0.3
Enter ".help" for instructions
sqlite> create table T(a INTEGER);
SQL error: database is locked
sqlite> SELECT name FROM sqlite_master
   ...> WHERE type='table'
   ...> ORDER BY name;
sqlite> .quit
 
 
Jeff Dever (B.C.S, E.Tech)
Software Developer
 
VAULT Technologies
Direct: 403.516.6316
Main: 403.723.2200
[EMAIL PROTECTED]
http://vaultpark.com  
 

Re: [sqlite] OT: Reply-To Munging Considered Useful

[David Morel]

Le ven 23/07/2004 à 14:38, Jay Macaulay a écrit :
> Removing the Reply-To makes me have to view the mail properties, find the
> From email, then type that in.  On the other hand, if I wanted to reply to
> the list before and not to a person directly, all I had to do was hilight
> the name and delete it.  A lot less painful than the other way around IMHO.

This is indeed all the more annoying as every other list I'm subscribed
to (thats makes about 2 dozens) behaves in a 'classic' way: hitting
'reply to' answers the poster, while 'reply to all' replies to the
poster and the list.

Sometimes, you don't want to pollute the list with an answer that has
qlready be given a billion times... such as the classic
smbfs-is-not-samba response that I send privately about 5 times a week
on the samba list -it's there in the archive alright but people don't
read the archives- which is about the only miserable piece of knowledge
I can share. I guess if pepole saw it all the time they would get sick
of me quite quickly. Still, I could help quite a few people with this
one.

So, what I mean is this: if it's harder to send a private response, You
can bet a lot of recurrent questions might end up unanswered.

-- 
***
[EMAIL PROTECTED]
OpenPGP public key: http://www.amakuru.net/dmorel.asc



signature.asc
Description: Ceci est une partie de message	=?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e=2E?=

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Matt Sergeant]

On Fri, 23 Jul 2004, Brass Tilde wrote:

> > Better:  Somebody please write me a simple, secure, mail handler to
> > replace qmail/ezmlm that lets each user decide for themselves whether
> > they want a Reply-To back to the mailing list or unmunged headers.
> >
> > I'll be happy to supply volunteers with a detailed specification of
> > what I am looking for in a mail system.
> 
> Funny you should mention that.  I've been looking for one of those myself,
> and had half decided to write my own.  Problem is, I don't know anything
> about mail processing, so it's going to be a long road.
> 
> If you don't mind though, I'd like to see those specs.  Even though I can't
> promise you anything, I'm interested in seeing other folks' ideas.  If you
> have them already put together, of course; I don't expect you to create them
> for me when I can't deliver what you're looking for.

http://siesta.sourceforge.net/


__
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
__

Re: [sqlite] not supported keywords

[Bert Verhees]

D. Richard Hipp wrote:
Bert Verhees wrote:
Hello,
Are there any plans in the near feature to implement the keywords
EXISTS and ALTER TABLE?
These to keywords/functions are very important when using SQLite in a 
MDA (Model Driven Architecture) where SQL-code is generated?

There are plans to add EXISTS and ALTER TABLE and a few other things.
Whether or not this occurs in the "near future" depends on your
definition of "near", I suppose.

I hope it will be soon.
I did some tests with Firebird (in embedded version). It is about half 
the speed in Delphi Architect.
Maybe it also depends on other things, that could be possible.

My application does not run smoothly with firebird, it does with SQLite.
But Firebird has support for these two keywords, so now I synchronize 
with a Firebird-database, and let the MDA-environment evolve and check 
the table.
Now I am looking to find my way back with the new database to SQLite, 
hope I will find a way.

But as you can see, for me, there is a big need for these commands.
It is just those two commands that are missing, for now at least, and I 
am working quit some time with this.

regards
Bert Verhees

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Brass Tilde]

> Better:  Somebody please write me a simple, secure, mail handler to
> replace qmail/ezmlm that lets each user decide for themselves whether
> they want a Reply-To back to the mailing list or unmunged headers.
>
> I'll be happy to supply volunteers with a detailed specification of
> what I am looking for in a mail system.

Funny you should mention that.  I've been looking for one of those myself,
and had half decided to write my own.  Problem is, I don't know anything
about mail processing, so it's going to be a long road.

If you don't mind though, I'd like to see those specs.  Even though I can't
promise you anything, I'm interested in seeing other folks' ideas.  If you
have them already put together, of course; I don't expect you to create them
for me when I can't deliver what you're looking for.

Re: [sqlite] Select string functions

[Fredrik Axelsson]

[EMAIL PROTECTED] wrote:
Fredrik Axelsson <[EMAIL PROTECTED]> writes:
 

I need to count the number of occurances of different characters in a string
in my database. So that I can select records where a string contains a number
of occurences of a certain character (independent of order). I'm not sure if
this is easy or hard to accomplish, but I'd like a few pointers if possible=)
   

A simplistic approach might go something like this:
If you wanted all strings that had 3 'f's (case insensitive) in it:
 SELECT string
   FROM table
   WHERE string LIKE "%f%f%f%";
For case sensitive, you could do:
 SELECT string
   FROM table
   WHERE string GLOB "*f*f*f*";
This doesn't count the number of occurrences of different characters, but it
does select records where a string contains a number of occurrences of a
certain character.  If you really need to count occurrences of different
characters, you could create an INSERT and UPDATE trigger that calls a
user-supplied function which counts characters and updates another column (or
26 columns, one per letter, if you like).
Derrell
 

Thanks Derrel!
It actually need to count the occurence of three groups of letters
occuring in three different columns, I was worried that storing this
information would make the size of the database grow unneccesarily, But
maybe the complexity of the searches will be an even bigger problem
later on if I don't.
Thanks for the info=)
/Fredrik

Re: [sqlite] OT: Reply-To Munging Considered Useful

[D. Richard Hipp]

Dan Keeley wrote:
Or, why dont we have a vote, and once it's decided, tuff luck thats how 
it stays? :)

Better:  Somebody please write me a simple, secure, mail handler to
replace qmail/ezmlm that lets each user decide for themselves whether
they want a Reply-To back to the mailing list or unmunged headers.
I'll be happy to supply volunteers with a detailed specification of
what I am looking for in a mail system.
--
D. Richard Hipp -- [EMAIL PROTECTED] -- 704.948.4565

Re: [sqlite] not supported keywords

[D. Richard Hipp]

Bert Verhees wrote:
Hello,
Are there any plans in the near feature to implement the keywords
EXISTS and ALTER TABLE?
These to keywords/functions are very important when using SQLite in a 
MDA (Model Driven Architecture) where SQL-code is generated?

There are plans to add EXISTS and ALTER TABLE and a few other things.
Whether or not this occurs in the "near future" depends on your
definition of "near", I suppose.
--
D. Richard Hipp -- [EMAIL PROTECTED] -- 704.948.4565

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Dan Keeley]

Or, why dont we have a vote, and once it's decided, tuff luck thats how it 
stays? :)

_
It's fast, it's easy and it's free. Get MSN Messenger today! 
http://www.msn.co.uk/messenger

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Dan Keeley]

> If you put an Reply-To Header in the mail, you make this options
> disfunctional. Now, both buttons do the same thing, they don't work as
> expected.
>
> Freedom is all about choice. Reply-To tries to disable the choice. Not
> very friendly, in my opinion.
>
> Please switch off Reply-To again.
>
> Thank you.
Why not specify both in the reply-to? That way you can remove the one you 
dont want??

_
Use MSN Messenger to send music and pics to your friends 
http://www.msn.co.uk/messenger

Re: [sqlite] OT: Reply-To Munging Considered Useful

[epankoke]

I, on the other hand, prefer to have a reply-to in a mailing list go to that mailing 
list.  For the rare occassion that I reply to someone personally on a mailing list, I 
don't mind the extra steps to get their email address.  I think what you're going to 
find with this topic is that you will make just as many people unhappy either way the 
reply-to ends up being set.  It's just a matter of personal preference.

--
Eric Pankoke
Founder / Lead Developer
Point Of Light Software
http://www.polsoftware.com/


> > If you put an Reply-To Header in the mail, you make this options
> > disfunctional. Now, both buttons do the same thing, they don't work as
> > expected.
> >
> > Freedom is all about choice. Reply-To tries to disable the choice. Not
> > very friendly, in my opinion.
> >
> > Please switch off Reply-To again.
> >
> > Thank you.
> 
> I'd have to agree with author here.  While I assume most of you don't reply
> to the person directly, I like replying to people personally, mostly to
> avoid sending people unneeded and unwanted amounts of mail, when a reply is
> specific to a problem.
> 
> Removing the Reply-To makes me have to view the mail properties, find the
> From email, then type that in.  On the other hand, if I wanted to reply to
> the list before and not to a person directly, all I had to do was hilight
> the name and delete it.  A lot less painful than the other way around IMHO.
> 
> Jay Macaulay
> 

Re: [sqlite] Versions 2.8.15 and 3.0.3 available

[Roger Reghin \(Duty/Sedes/CWB\)]

I'll try it and give you a feedback!!

Thank you so much! =)

Roger.


- Original Message - 
From: "Andy Colson" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Thursday, July 22, 2004 9:56 PM
Subject: Re: [sqlite] Versions 2.8.15 and 3.0.3 available


> Roger Reghin (Duty/Sedes/CWB) wrote:
> > D. Hipp,
> >
> > is there a 2.8.15 pre-compiled DLL available??  I still couldn't find
any =(
> >
> > Thanks,
> >
> > Roger Reghin.
>
> I have compiled the 3.0.3 DLL as well as 2.8.15 (my first time with the
> 2.8.x so hopefully its right) and put them on my website:
>
> http://squeakycode.net/
>
> if anyone has problems please let me know.
>
> -Andy
>
> PS  I included the sqlite_encode_binary and decode_binary into the DLL,
> it just seemed like a good idea...  hope it wont break anything.
>

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Brass Tilde]

> As an experiment, I have turned on Reply-To munging.  Reply-To now
> points back to the mailing list (assuming I did it right).

I'll be interested in seeing how this turns out.  I've been waffling on the
issue since reading the original essay, but the rebuttal makes some good
points.  The purpose of a discussion list is to hold a discussion with a
group of people, not individuals.  For the majority of members, sending to
an individual will likely be the exception, not the rule.

I don't find it particularly onerous to copy the email address from the
message text (since my mailer kindly retrieves the *author's* email in my
reply text) and copy it to the "To" field.

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Jay Macaulay]

> If you put an Reply-To Header in the mail, you make this options
> disfunctional. Now, both buttons do the same thing, they don't work as
> expected.
>
> Freedom is all about choice. Reply-To tries to disable the choice. Not
> very friendly, in my opinion.
>
> Please switch off Reply-To again.
>
> Thank you.

I'd have to agree with author here.  While I assume most of you don't reply
to the person directly, I like replying to people personally, mostly to
avoid sending people unneeded and unwanted amounts of mail, when a reply is
specific to a problem.

Removing the Reply-To makes me have to view the mail properties, find the
>From email, then type that in.  On the other hand, if I wanted to reply to
the list before and not to a person directly, all I had to do was hilight
the name and delete it.  A lot less painful than the other way around IMHO.

Jay Macaulay

Re: [sqlite] OT: Reply-To Munging Considered Useful

[Michael Roth]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
D. Richard Hipp wrote:
| As an experiment, I have turned on Reply-To munging.  Reply-To now
| points back to the mailing list (assuming I did it right).
IMHO this is a annoying experiment.
MUAs have two buttons or key-bindings: One replies to the author, and
one replies to all.
If you put an Reply-To Header in the mail, you make this options
disfunctional. Now, both buttons do the same thing, they don't work as
expected.
Freedom is all about choice. Reply-To tries to disable the choice. Not
very friendly, in my opinion.
Please switch off Reply-To again.
Thank you.
-BEGIN PGP SIGNATURE-
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
iD8DBQFBAQFsSIrOxc3jOmoRAp0GAJ9Epo7qO1yTQ8uDYwWTBmwGDWUDWwCdEtTu
tbdY6WNx63fzC3gUkqE0Uzg=
=vn3g
-END PGP SIGNATURE-

Re: [sqlite] very large database file

[Christian Smith]

On Thu, 22 Jul 2004, Will Leshner wrote:

>Ok. I think I know what is going on. Each of the rows in my database
>is "too big" to fit on the main page (about 320 bytes for each row).
>So each one is forcing an overlow page. I'm assuming that each
>overflow page only stores one row of data. In other words, the
>overflow pages don't share rows as the main pages do. So since my rows
>are just a little too big to fit on the main page, most of each
>overflow page is unused. And since just about every one of my 80,000
>rows is creating a 1k overflow page that is largely unused, I'm ending
>up with a very large database file that is made up mostly of unused
>space.
>
>Does that sound about right?

That'll be right.

You might want to try SQLite 3, as 3.0.3 has variable page size enabled,
allowing you to use bigger pages which can store larger rows without
overflowing[0]. To do the same with SQLite 2 would require a recompile and
incompatible file with default SQLite 2 files.

Christian

[0] Each btree page can hold at least 4 cells. Increase page size, and
cell size goes up.

-- 
/"\
\ /ASCII RIBBON CAMPAIGN - AGAINST HTML MAIL
 X   - AGAINST MS ATTACHMENTS
/ \

Re: [sqlite] OT: Reply-To Munging Considered Useful

[D. Richard Hipp]

Nuno Lucas wrote:
Don't want to start any flame war, but found this, which I consider are 
valid points for the "reply-to" mail header:

http://www.metasystema.net/essays/reply-to.mhtml
Written in response to Chip Rosenthal's "Reply-To Munging Considered 
Harmful" (http://www.unicom.com/pw/reply-to-harmful.html).

As an experiment, I have turned on Reply-To munging.  Reply-To now
points back to the mailing list (assuming I did it right).
I agree that the default Reply-To back to the original author is
very annoying.  99% of the time you want to reply back to the
list.  On those rare occasions when you want to reply to just
the author, it isn't too much trouble to type in an email address.
If, on the other hand, we start running into problems with
vacation replies and other kinds of autoresponder, I'll go
back to the non-munged Reply-To.
--
D. Richard Hipp -- [EMAIL PROTECTED] -- 704.948.4565

Re: [sqlite] Please confirm this is a bug !

[D. Richard Hipp]

Andrew Clark wrote:
Not knowing the code that well but guessing I would say that because the 
statements are on a single line I believe maybe the SQL interpreter is 
reading the BEGIN (start a transaction) but not reading the rest of the 
commands when an error occurs and therefore doesn't see the COMMIT (end 
transaction) at the end.

That is correct.  When you have multiple statements in a single sqlite3_exec()
and an error is encountered in one of the statements in the middle, none
of the statements that follow are executed.  This is by design.
I find it useful to execute the COMMIT using a separate sqlite3_exec().
If the first sqlite3_exec() returns successfully, then execute COMMIT
in a separate sqlite3_exec(), otherwise issue a ROLLBACK.
--
D. Richard Hipp -- [EMAIL PROTECTED] -- 704.948.4565

[sqlite] Please confirm this is a bug !

[Andrew Clark]

Hi,
Using the C API if i submit the statements:
BEGIN;
CREATE TABLE test (id, name);
COMMIT;
within a single Exec command (or as a single line on the command line 
tool) then it works the first time and displays an error the second time 
as expected due to the table already existing.

However, when the error occurs I am then unable to run anymore 
transactions because it says that a transaction is already active.

So if I ran that set of statements a third time it would come up with 
the error:

SQL error: cannot start a transaction within a transaction
When reading the documentation I realised that sending a set of 
statements in an Exec command is treated as a single transaction
anyway and therefore I can remove the BEGIN and COMMIT statements.

When I do this then I can run that set of statements multiple times 
without SQLite displaying an error saying that a Transaction is already 
active.

I haven't looked deep into the code but I thought I would ask the list 
if this is a bug or if it's just me not reading the documentation 
properly.

Not knowing the code that well but guessing I would say that because the 
statements are on a single line I believe maybe the SQL interpreter is 
reading the BEGIN (start a transaction) but not reading the rest of the 
commands when an error occurs and therefore doesn't see the COMMIT (end 
transaction) at the end.

Andrew
--
Andrew Clark

[sqlite] OT: Reply-To Munging Considered Useful

[Nuno Lucas]

Don't want to start any flame war, but found this, which I consider are 
valid points for the "reply-to" mail header:

http://www.metasystema.net/essays/reply-to.mhtml
Written in response to Chip Rosenthal's "Reply-To Munging Considered 
Harmful" (http://www.unicom.com/pw/reply-to-harmful.html).

Just food for the brain ;)
Regards,
~Nuno Lucas

[sqlite] not supported keywords

[Bert Verhees]

Hello,
Are there any plans in the near feature to implement the keywords
EXISTS and ALTER TABLE?
These to keywords/functions are very important when using SQLite in a 
MDA (Model Driven Architecture) where SQL-code is generated?

thanks
Bert Verhees