PKs or any other constraints. Newer versions of SQLite will attempt
to preserve the column affinities, but, IIRC, older version will not
(and all columns will have a NONE affinity).
CREATE TABLE...AS SELECT is great for temp tables, but it is not a
useful way to truly "copy" a
g:8080/cgi-bin/mailman/listinfo/sqlite-users
> > >
> >
> >
> >
> > --
> > D. Richard Hipp
> > d...@sqlite.org
> > _______
> > sqlite-users mailing list
> > sqlite-users@sqlite.org
> > http://sqlite.org:80
#x27;re going to see seeks and random rotation delays that
should be even larger than this.
-j
> sort fsync.dat | tail -n 5
> fsync time: 0.0025
> fsync time: 0.0025
> fsync time: 0.0026
> fsync time: 0.0027
> fsync time: 0.0059
>
--
Jay A. Kreibich < J A Y @ K R E
e, but it isn't doing anything.
If you're actually writing data to a spinning disk, I would expect
times to be varied, and average out to slightly more than half a
rotation. And rotations still take a LONG time in modern computing
terms.
-j
--
Jay A. Kreibich < J A Y @ K R
ways distinct.
> Is there a way I could achieve this?
You want the DISTINCT on the aggregate, not the result set:
SELECT group_concat( DISTINCT TB.F2, ' ' ) AS
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is im
omplex than it first
sounds.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
feel uncomfortable." -- Angela Johnson
likely source-control systems. And
they're typically very complex, traditionally have a central "master"
dept, and often require human intervention to untangle the mess when
a merge goes wrong. They're actually an easy case.
-j
--
On Fri, Jun 15, 2012 at 12:12:22PM +, Black, Michael (IS) scratched on the
wall:
> Could the SQLite code add a new dummy function for a callback which
> we can then override with our own library?
Just use an SQL function:
http://sqlite.org/c3ref/create_function.html
-j
--
where aph>0 or hctl=1" That will also
avoid duplicate rows if a specific row meets both conditions.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
by the shell and many programs.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
feel uncomfortable." -- Angela Johnson
__
ntf conversion.
By default, it is limited to 6 decimal places. If it only sees
zeros, it will roll the number back up. The number isn't "wrong", it
is just being displayed in a shortened notation.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence
ow
> where it is getting this value from however.
If by "this value" you mean the 6, that's the default precision for
"%f" conversion in the printf() family of functions.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like under
do not have to escape a * inside a parameter, you cannot
use a parameter to reference the name of a table or column.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has
does not take databases as table names. "test" works
> as an argument, but "main.test" doesn't.
No, but following the standard of every other PRAGMA, this does work:
PRAGMA main.table_info( test );
See the syntax diagrams here: http://sqlite.org/pra
from the index itself, and will
never touch the main table record. This is more efficient, even if
the index is known to have poor diversity.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it t
()
or _finalize() because of locking semantics. If your program can call
one of those APIs and maintain correct flow, it should be calling them.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showin
-interaction. Even then, it might not
work... I have no idea if things are "inserted" into sqlite_master in
the standard way.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to
how to use them. It happens to be the longest
chapter in the book, and walks through two full examples. One of the
examples is using a virtual table to map web-server logs to an SQLite
table without importing the data. From what you said, that sounds
somewhat similar to your problem.
cing them to cast back and forth for the exception to the
rule seems like a bit of waste.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
fee
t;
> How can this be fixed ?
Move two timezones to the west.
(By default all times and dates are UTC.)
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tenden
On Wed, Jun 27, 2012 at 05:45:41PM +0200, deltagam...@gmx.net scratched on the
wall:
> Am 27.06.2012 17:40, schrieb Jay A. Kreibich:
> >On Wed, Jun 27, 2012 at 05:37:55PM +0200, deltagam...@gmx.net scratched on
> >the wall:
> >
> >>Hello,
> >>
>
v2() deals with this condition
automatically is generally considered to be a feature.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the t
pe. As long as you use the encoder function
for inputs and the decoder for all outputs, you should be good. That
starts to get deep into your SQL, however. The ability to define
native types is similar in complexity to adding user-defined
functions.
Just a thought. Any opinions?
-j
d, but there is a hard
limit of (2^31 - 1), or 2GB.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
feel uncomforta
L rows, you need to
update only the non-NULL rows:
UPDATE table SET column=NULL WHERE column IS NOT NULL;
As for sqlite3_changes() returning 0, that doesn't sound right unless
you're checking inside the trigger.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H
doesn't change what it does.
There is, however, little argument that the trigger is doing exactly
what one would expect. You are applying an update operation to every
row, and the trigger is firing for every row.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Inte
t;>?3.
> (passing a null parameter to the above won't even work!)
Well, no, it won't, because you're using the wrong operator.
Use "WHERE col1 IS NOT ?1 AND..." and it all works fine.
> No surprises there. Oracle has never managed to impress me.
de-effect of the fact that CREATE statements are copied
into the sqlite_master table as literals, and not re-written? (Is
that even true?)
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to
name );
sqlite3_prepare_v2( db, sql_str, -1, &stmt, NULL );
sqlite3_free( sql_str );
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to ma
orts several extensions to the
standard printf() syntax in the sqlite3_*printf() family of
functions. Both %q and %Q can be used for values, while %w can be
used for identifiers. The sqlite3_*printf() functions will properly
quote and sanitize any value in the generated string.
There is also
On Wed, Jul 04, 2012 at 08:29:33AM -0500, Jay A. Kreibich scratched on the wall:
> On Wed, Jul 04, 2012 at 01:09:01AM -0500, Nico Williams scratched on the wall:
>
> > But this would
> > just be a glorified (if safer) variant of sqlite3_mprintf() -- for
> > apps that allow
uality:
http://www.sqlite.org/contrib/
Also see:
sqlite3_auto_extension()
http://www.sqlite.org/c3ref/auto_extension.html
sqlite3_load_extension()
http://www.sqlite.org/c3ref/load_extension.html
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is
nce boost from a VACUUM. I'd be more concerned
about filesystem fragmentation than I would be about SQLite
fragmentation.
> You could use the shell tool to turn the database file into SQL commands,
> and then back into a new database file on disk. This will both
> defragment t
the latest version of SQLite, otherwise it won't work.
Also, not to state the obvious, but you can only share a :memory:
database across connections that originate from the same process.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is i
ost welcome
> >
> >Regards
> >Olivier
> >___
> >sqlite-users mailing list
> >sqlite-users@sqlite.org
> >http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
> >
> >
>
>
>
ering the occasional spam message.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
feel uncomfortable." -- Angela Johnson
___
ect
>
> Single table with 20 columns. Unless your data is very unusual.
However, insert/updates/deletes are likely to be faster on the
smaller tables.
Worry about design first, then optimize for speed. "Normalize 'till
it hurts, denormalize until it works."
-j
really something of an error condition. It means your application
lost track of something, and failed to free a dependent resource.
Having your object blindly finalize statements is very likely to
leave a dangling pointer elsewhere in the application.
-j
--
Jay A. Kreibich < J A Y
lled, but, yes... this use of SQLITE_STATIC is
acceptable (and somewhat common).
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
feel unc
would have to be _finalize() or _clear_bindings().
But yes... the key is that the memory remains valid for the lifetime
of the binding, not the fact that is or isn't statically allocated.
-j
> -Original Message-
> From: sqlite-users-boun...@sqlite.org
> [mailto:sqlite
ing SQLite":
http://shop.oreilly.com/product/9780596521196.do
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendenc
yped languages, it might be better to expose the
sqlite3_value object as an opaque object and provide the standard
bind, column, result, and value APIs. You don't have a lot of choice
in that case. Just be aware there is no way to create an
sqlite3_value object from scratch. You
an issue, and you don't want hundreds of
connections banging on the same file, but that's true no matter if
the connections come from the same process or not.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you
On Sat, Sep 01, 2012 at 11:56:33PM -0700, J Decker scratched on the wall:
> On Sat, Sep 1, 2012 at 8:24 PM, Jay A. Kreibich wrote:
> > On Sat, Sep 01, 2012 at 07:37:04PM -0700, J Decker scratched on the wall:
> >> On Sat, Sep 1, 2012 at 7:32 PM, Simon Slavin wrote:
> >
(though non-zero)
chance that a power failure at just the wrong time could corrupt
the database in NORMAL mode. But in practice, you are more likely
to suffer a catastrophic disk failure or some other unrecoverable
hardware fault.
-j
--
Jay A. Kreibich < J A Y @
s describe
this fairly well, but from the sound of it you need FULL for
durability. On the other hand, WAL requires fewer write to commit a
transaction, so (if I'm reading this correctly) FULL in WAL mode is
much faster than FULL in non-WAL mode.
-j
--
Jay A. Kreibich < J A Y @
On Mon, Sep 10, 2012 at 09:50:58PM -0500, Jay A. Kreibich scratched on the wall:
> On Tue, Sep 11, 2012 at 01:58:23AM +0100, Simon Slavin scratched on the wall:
> > On 11 Sep 2012, at 12:55am, Keith Chew wrote:
>
> > > and I know FULL (1) will provide that. The question i
le, in SQL "NULL OR 1" is 1
(true) and "NULL AND 0" is 0 (false).
Arguments about the semantic details of Relational algebra aside, if
you treat NULL as "unknown", most of the database operators and
functions make sense.
http://en.wikipedia.org/wiki/Three-value
takes a bit to wrap your head around, but don't
worry about it too much. Unless you're writing a VT that provides a
specialized index, you can usually just ignore it and get the basic
VT working with table scans before you worry about making the VT
index aware. A lot of the VT mod
o this kind of thing in a traditional database.
Did it make my life easier, the code simpler, and the database
smaller and more compact? Heck, yes.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but
y wants strong typing, that's easy
enough to do. Just add a check constraint to your column defs:
CREATE TABLE t (
i integer CHECK ( typeof( i ) == 'integer' ),
t text CHECK ( typeof( t ) == 'text' ),
r float CHECK ( typeof( r ) ==
On Mon, Sep 24, 2012 at 09:05:51PM +0400, Yuriy Kaminskiy scratched on the wall:
> Jay A. Kreibich wrote:
> > And finally, for anyone that really wants strong typing, that's easy
> > enough to do. Just add a check constraint to your column defs:
> >
> &
clean /tmp on reboot, and that can be months, if not years, on many
Unix systems. Some don't clean /tmp at all.
The "create and unlink" pattern is so common, many UNIX systems have
a tmpfile() or similar library call to do the whole thing... create a
unique file in /tm
this will not solve every problem. Even with a
timeout, there are situations when you can still get a locking error
and your only choice is to rollback and try again.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you ha
nd handle the vast, vast majority of SQLITE_BUSY errors,
however.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make
pplication terminates for any reason. It will create the file
in the system's default temp space, which is /tmp in the case of UNIX
systems. This call is part of the POSIX platform standard, as well
as the ISO C-90 standard. I'm sure one could trace its roots back
pretty far into th
on that includes a concat() function. That way you get the
function you want and the behavior you want.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tende
the table, the whole application will simply crash. Your fault.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
feel uncomfortable." -- An
#x27;t work.
SQL string literals use single quotes.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
feel uncomfortable." -- Angela Johnson
les ?
Not simpler, but cleaner... write a VFS plugin that reads/writes to a
memory block. Use the backup API to go straight in and out of that,
rather then a file.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that
te, all columns are in all indexes even if the column contains a
> NULL. NULL has a sorting order, and anything that does
Rows, Simon, rows... not columns. Watch your terminology or your
answers will be more confusing than the questions.
-j
--
Jay A. Kreibich < J A Y @ K R E I
rows, it will slow down, rather than speed up, a
query.
Indexes are not magic bullets, and using them properly requires
understanding how they work and how they are used.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important th
On Sun, Nov 25, 2012 at 12:41:21PM -0700, Keith Medcalf scratched on the wall:
> > On Sunday, 25 November, 2012, 11:58, Jay A. Kreibich wrote:
>
> > each column is usually undesirable. A given SELECT can usually only
> > use one index per query (or sub-query), so it
I would not assume the backup API writes the file front to back,
especially if the database is modified while the backup is taking
place.
A custom VFS that just "writes" the file to a big chunk of memory
makes the most sense.
-j
--
Jay A. Kreibich < J A Y @ K R E I
backup gets caught in a
restart loop, some people choose to make the backup a more atomic
operation by having the backup "step" function copy all the pages in
one go. In that case it is likely that the majority of pages are
written out in-order, but I wouldn't want to bank on
to setup an FK that
references a column or set of columns that does not have a UNIQUE
constraint, either the FK is broken or the parent table is broken.
...which is not to say a general purpose tool still needs to deal
with this, as there are plenty of broken database designs out there.
most
systems the default page size is 1K, but it can be 4K on some Windows
systems. That makes the 16MB look a shade big, but it might be about
right if you're running on a Windows system, or if you've adjusted
the default page size and/or cache size.
-j
--
Jay A.
thod is to edit sqlite_master directly.
> I know it is all saved as text
No, it isn't. That was true of SQLite 2, but SQLite 3 stores types
in their native format.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important th
On Sun, Dec 02, 2012 at 12:52:33PM -0800, Igor Korot scratched on the wall:
> Jay,
>
> On Sun, Dec 2, 2012 at 12:16 PM, Jay A. Kreibich wrote:
> > On Sun, Dec 02, 2012 at 11:58:54AM -0800, Igor Korot scratched on the wall:
> >> Hi, ALL,
> >> ALTER TABLE comm
e when rows have been deleted
from the database, however.
Not to ask the obvious, but are you sure the rows were actually
deleted? Was auto-vacuuming on?
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
bu
ed-- or the whole statement should be considered
invalid and an error thrown.
While the phantom parameter issue might be worth addressing, in
this specific case I think it is fair to call the query incorrect.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"
uot;very large but not exactly known" number, minus one of the same,
might result in anything from +Inf to -Inf. Hence, the "unknown" NULL.
I'd guess most IEEE 754 results of NaN will produce a NULL in SQL,
since both can mean "unknown" or "missing/illog
es together into a single, less clear value, just
for the sake of making one query, rather than two? Or even one query,
but with an extra line or two of code in the parse function?
Why not just deal with values in their native, and more correct,
"list of images" format?
-j
in a data storage system. Write a C or assembly program and have all
the close, fine-grain detail you want. As you've pointed out, SQLite
is more than capable of storing and retrieving non-numeric IEEE 754
values, so it is doing its core job just fine.
-j
--
Jay A. Kreibich &
doesn't
deal with floating point numbers for your definition of "as it should."
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
f
t; retrieving non-numeric IEEE 754 values
>
> No, it doesn't support storing and retrieving NaNs.
You're right... Even using the C API, if you try to call
sqlite3_bind_double( stmt, p, 0.0 / 0.0 ), the NaN is converted into
a NULL before it is written to the database. Inf works
--
Finally, you can call this in your application:
sqlite3_auto_extension( (void(*)(void))extension_functions_init );
With that done, any database you open with that application should
have the extens
t; range.
** In my own experiences working at NCSA, two national labs, and
several university HPC projects, most people doing "scientific
computing" don't have a damn clue. Most of the "scientific
computing" code I saw was written by domain grad students or
ary requires no
code changes.
The dlopen() and related functions are for application controlled
linking. They're like the LoadLibrary() functions under Windows.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you
tes in the header file.
src/test_intarray.h
http://www.sqlite.org/cgi/src/artifact/489edb9068bb926583445cb02589344961054207
src/test_intarray.c
http://www.sqlite.org/cgi/src/artifact/d879bbf8e4ce085ab966d1f3c896a7c8b4f5fc99
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"I
al table that does something
> pretty much like the above. We've recently started using it some for
> our test cases.
Ooo... that's even more nifty and simple.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you
>
> .tables T
>
> Produce no results.
You can't. The CLI dot-commands only show results for the
"main" and "temp" databases.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important th
f/busy_handler.html <= deadlock info
http://sqlite.org/c3ref/busy_timeout.html
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
feel uncomfortable.&q
d release all the locks. Of course, many, many
applications out there don't deal with this correctly, so you would
need to be careful with a general-access database.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that
ht try changing the journal mode to
"truncate."
http://www.sqlite.org/pragma.html#pragma_journal_mode
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people
cgi-bin/mailman/listinfo/sqlite-users
> >>>
> >> ___
> >> sqlite-users mailing list
> >> sqlite-users@sqlite.org
> >> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
> >>
> >
here MyField isnull is also supported.
http://www.sqlite.org/lang_expr.html ("is null" is just a standard
"is" with the right side expression being a literal NULL.)
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear:
ngs. If you're
using an extensive number of OMIT flags, a bit of code clean-up may be
required.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people ha
he query syntax can get quite messy.
I would put together a few example cases of what you're trying to do
with your application. In addition to the data layout, pay specific
attention to the types of queries you need to run and how you're
going to set those up.
-j
--
Jay A.
le associated with the
transaction. Make sure your application has write/create permissions
to the directory with the database file. If you cannot provide that,
put the database in a subdirectory and provide the permissions on the
subdirectory.
-j
--
Jay A. Kreibich < J A Y @ K R
or
both storage space and your specific access patterns.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has the tendency to make them
feel uncomfortable." -- Angela
S ( 2, 2, 22, 222 );
sqlite>
sqlite> SELECT coalesce( over.a, real.a ) AS a,
...>coalesce( over.b, real.b ) AS b,
...>coalesce( over.c, real.c ) AS c
...> FROM real LEFT OUTER JOIN over USING ( id );
1|10|100
2|22|222
3|30|300
sqlite>
The order of the j
e. Using the
OUTER JOIN allows you to use arbitrary WHERE clauses, including
queries that return multiple rows. That said, most people would
likely consider the UNION approach easier to understand.
-j
>
> Keith
>
> On Jul 20, 2011 11:35am, "Jay A. Kreibich" wrote
;
> 61311;18461F;
In SQL terms, these are the exact same. Table rows are *unordered*
and can be returned in any order the database wants.
If you need a result in a specific order, you must use an ORDER BY
clause in your SELECT statement.
-j
--
Jay A. Kreibich < J A Y
It is the application code that needs to get fixed to accept
data in an order that the SQL standard can provide. This isn't true
of only SQLite, but all SQL database systems.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is impor
ry. in my
> programs main healer i did : #include "sqlite3ext.h" and linked the
> LIB file.
You want to use "sqlite3.h" in applications.
The "sqlite3ext.h" file is for building library extensions
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
create index idx2 on t(i,rowid);
> Error: table t has no column named rowid
>
> Any particular reason it can't be included in an index?
Because it is always included as the last column.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like u
On Thu, Jul 28, 2011 at 12:57:45PM -0700, km4hr scratched on the wall:
>
> Is there a way to edit SQL commands typed in on the command line?
Only if the "sqlite3" tool is compiled with a copy of the readline
library. In that case, just hit up-arrow.
-j
--
Jay A.
t; to the database files.
"PRAGMA locking_mode = EXCLUSIVE" never releases the locks.
http://www.sqlite.org/pragma.html#pragma_locking_mode
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but
e access?
You can set the file permissions. That's a more appropriate means to
prevent this kind of operation.
-j
--
Jay A. Kreibich < J A Y @ K R E I B I.C H >
"Intelligence is like underwear: it is important that you have it,
but showing it to the wrong people has
801 - 900 of 961 matches
Mail list logo