Brannon King wrote:
> [...] I'd like to propose that we upgrade to something more modern and
> secure like Sympa or mlmmj, or even a more drastic system upgrade to
> something like Redmine -- a project tracker + forum.
I would propose instead to upgrade to a NNTP. (I even wrote a NNTP software
On 10/13/19, Shawn Wagner wrote:
> The documentation for a column with NUMERIC affinity says
>
>> When text data is inserted into a NUMERIC column, the storage class of
> the text is converted to INTEGER or REAL (in order of preference) if such
> conversion is lossless and reversible.
>
Thank
On 10/13/2019 8:04 PM, Shawn Wagner wrote:
That's what I told the guy having the original issue to do, yes. That's not
important.
My concern is why a conversion that's only supposed to happen if it's
lossless is in fact happening and causing data loss.
You define the term "lossless"
That's what I told the guy having the original issue to do, yes. That's not
important.
My concern is why a conversion that's only supposed to happen if it's
lossless is in fact happening and causing data loss.
On Sun, Oct 13, 2019, 4:48 PM Igor Tandetnik wrote:
> On 10/13/2019 7:25 PM, Shawn
On 10/13/2019 7:25 PM, Shawn Wagner wrote:
I wouldn't call that conversion, or any other, lossless unless it can be
turned back into a string that's character for character identical with the
one that was originally inserted.
If you want the text preserved character for character, store it in
sqlite> create table x(x numeric);
sqlite> insert into x values ('0012');
sqlite> select typeof(x), x from x;
integer|12
sqlite> select printf('%04d', x) from x;
0012
Presentation is a user/application problem. Not a database data problem.
--
The fact that there's a Highway to Hell but only a
On October 13, 2019 7:25:50 p.m. EDT, Shawn Wagner
wrote:
>I wouldn't call that conversion, or any other, lossless unless it can
>be
>turned back into a string that's character for character identical with
>the
>one that was originally inserted.
It's lossless if you consider the input as the
Yes, I know. (This all comes from someone on stack overflow who used
"string" as a column type and thus ran into this issue because that of
course results in numeric affinity)
On Sun, Oct 13, 2019, 4:27 PM Simon Slavin wrote:
> On 13 Oct 2019, at 10:11pm, Shawn Wagner wrote:
>
> > As you can
On 13 Oct 2019, at 10:11pm, Shawn Wagner wrote:
> As you can see, the leading zeros in the original string are gone and it's
> been converted to an integer. This seems to violate the "lossless and
> reversible" constraint. Shouldn't it be kept as text?
You defined the column as 'numeric'.
I wouldn't call that conversion, or any other, lossless unless it can be
turned back into a string that's character for character identical with the
one that was originally inserted.
On Sun, Oct 13, 2019, 4:10 PM Igor Tandetnik wrote:
> On 10/13/2019 5:11 PM, Shawn Wagner wrote:
> > The
On Sunday, 13 October, 2019 15:23, Adam Levy wrote:
>My application requires a way to build a "pending state" in the database
>while allowing users to query data from both the "official state" and the
>"pending state". I am achieving this using sessions and snapshots.
SQLite3 does not do
On 13 Oct 2019, at 10:23pm, Adam Levy wrote:
> My application requires a way to build a "pending state" in the database
> while allowing users to query data from both the "official state" and the
> "pending state". I am achieving this using sessions and snapshots.
You are depending on finicky
On 10/13/2019 5:11 PM, Shawn Wagner wrote:
The documentation for a column with NUMERIC affinity says
When text data is inserted into a NUMERIC column, the storage class of
the text is converted to INTEGER or REAL (in order of preference) if such
conversion is lossless and reversible.
3.30, though it goes back to at least 3.8.7 (the oldest version I have
available to test with)
On Sun, Oct 13, 2019, 3:03 PM Igor Korot wrote:
> Hi,
>
> On Sun, Oct 13, 2019 at 4:12 PM Shawn Wagner
> wrote:
> >
> > The documentation for a column with NUMERIC affinity says
> >
> > > When text
Hi,
On Sun, Oct 13, 2019 at 4:12 PM Shawn Wagner wrote:
>
> The documentation for a column with NUMERIC affinity says
>
> > When text data is inserted into a NUMERIC column, the storage class of
> the text is converted to INTEGER or REAL (in order of preference) if such
> conversion is lossless
My application requires a way to build a "pending state" in the database
while allowing users to query data from both the "official state" and the
"pending state". I am achieving this using sessions and snapshots.
When pending data first comes into the application, I take a snapshot of
the
The documentation for a column with NUMERIC affinity says
> When text data is inserted into a NUMERIC column, the storage class of
the text is converted to INTEGER or REAL (in order of preference) if such
conversion is lossless and reversible.
But consider:
sqlite> create table foo(bar
On Sunday, 13 October, 2019 01:41, Adam Levy wrote:
>One of the requirements for sqlite3_snapshot_get is that "One or more
>transactions must have been written to the current wal file since it was
>created on disk (by any connection). This means that a snapshot cannot be
>taken on a wal mode
One of the requirements for sqlite3_snapshot_get is that "One or more
transactions must have been written to the current wal file since it was
created on disk (by any connection). This means that a snapshot cannot be
taken on a wal mode database with no wal file immediately after it is first
19 matches
Mail list logo