Did you consider using wiredtiger?
Le lun. 19 nov. 2018 à 11:47, Danny Milosavljevic
a écrit :
> Hi Ludo,
>
> >It doesn’t seem to help much, perhaps because the query is too complex?
>
> Yeah, probably.
>
> According to the docs a log message is supposed to appear when it is doing
> it.
>
> We
Hi Ludo,
>It doesn’t seem to help much, perhaps because the query is too complex?
Yeah, probably.
According to the docs a log message is supposed to appear when it is doing it.
We should just special-case the common queries so the optimizer has a
easier life.
pgp8qd5F8FXLx.pgp
Description:
Hi
On 2018-11-10 18:33, Ludovic Courtès wrote:
snip
--8<---cut here---start->8---
sqlite> CREATE INDEX Builds_index_evaluation ON Builds(evaluation);
snip
Anything else we should do?
I woke up today with an idea. :)
Now on berlin it seems that all
On Tue, 13 Nov 2018 09:10:30 +0100
Clément Lassieur wrote:
> Hi Björn,
>
> Björn Höfling writes:
>
> > We are directly relying on the rowid here, there is no explicit
> > id-column.
> >
> > This could lead to unpredicted results and reorderings (6th Quirk in
> > document):
> >
> >
On Tue, 13 Nov 2018 00:27:15 +0100
Danny Milosavljevic wrote:
> Hi Björn,
>
> On Mon, 12 Nov 2018 19:50:44 +0100
> Björn Höfling wrote:
>
> > Hm. This code smells ... It looks too complicated.
>
> I was trying to cut down the number of prepared statements in use and
> prevent a
Hello,
Danny Milosavljevic skribis:
> On Mon, 12 Nov 2018 19:50:44 +0100
> Björn Höfling wrote:
>
>> Hm. This code smells ... It looks too complicated.
>
> I was trying to cut down the number of prepared statements in use and prevent
> a
> combinatorial explosion while keeping the kinds of
Hi,
Danny Milosavljevic skribis:
> On Sun, 11 Nov 2018 18:06:00 +0100
> l...@gnu.org (Ludovic Courtès) wrote:
>
>> I don’t really know what additional index to create (and I’d rather let
>> SQLite do it for me, if it were possible).
>
> Not exactly what you mean but there's this:
>
>
Hi Björn,
Björn Höfling writes:
> We are directly relying on the rowid here, there is no explicit
> id-column.
>
> This could lead to unpredicted results and reorderings (6th Quirk in
> document):
>
> https://www.sqlite.org/rowidtable.html
>
> We should add a column:
>
> id INTEGER PRIMARY KEY
On Tue, 13 Nov 2018 00:31:40 +0100
Danny Milosavljevic wrote:
> Hi Ludo,
>
> On Sun, 11 Nov 2018 18:06:00 +0100
> l...@gnu.org (Ludovic Courtès) wrote:
>
> > I don’t really know what additional index to create (and I’d rather let
> > SQLite do it for me, if it were possible).
>
> Not
Hi Ludo,
On Sun, 11 Nov 2018 18:06:00 +0100
l...@gnu.org (Ludovic Courtès) wrote:
> I don’t really know what additional index to create (and I’d rather let
> SQLite do it for me, if it were possible).
Not exactly what you mean but there's this:
https://www.sqlite.org/lang_analyze.html
It does
Hi Björn,
On Mon, 12 Nov 2018 19:50:44 +0100
Björn Höfling wrote:
> Hm. This code smells ... It looks too complicated.
I was trying to cut down the number of prepared statements in use and prevent a
combinatorial explosion while keeping the kinds of queries we can do open.
Either the value of
Hello all,
Le lun. 12 nov. 2018 à 19:51, Björn Höfling <
bjoern.hoefl...@bjoernhoefling.de> a écrit :
> Hi Ludo,
>
> On Sun, 11 Nov 2018 18:06:00 +0100
> l...@gnu.org (Ludovic Courtès) wrote:
>
> > Indeed, that solves the problem for this simple example, thanks!
> >
> > Now, if I go back to the
Hi Ludo,
On Sun, 11 Nov 2018 18:06:00 +0100
l...@gnu.org (Ludovic Courtès) wrote:
> Indeed, that solves the problem for this simple example, thanks!
>
> Now, if I go back to the big query that /api/latestbuilds makes¹, the
> result is still pretty bad:
>
> --8<---cut
Hi Björn,
Björn Höfling skribis:
> The link you provided explains it: The column over which you are sorting
> (stoptime) is not indexed. Add it to the (same) index:
>
> sqlite> DROP INDEX Builds_index_evaluation;
> sqlite> CREATE INDEX Builds_index_evaluation ON Builds(evaluation, stoptime);
>
On Sat, 10 Nov 2018 18:33:23 +0100
l...@gnu.org (Ludovic Courtès) wrote:
> Now, ‘db-get-builds’ in Cuirass uses a more complex query. In
> particular, it orders things, very roughly along these lines:
>
> --8<---cut here---start->8---
> sqlite> EXPLAIN QUERY
Hello!
I was investigating the slowness of our /api/latestbuilds requests on
berlin.
I found that if we have just the two indexes currently defined in
‘schema.sql’, basically everything involves a table scan:
--8<---cut here---start->8---
sqlite> EXPLAIN
16 matches
Mail list logo
