I'm not involved in Entropy's development but I think using a full-blown
database for such a task is overkill. Note that by overkill I don't mean
useless, I rather mean suboptimal. Take Postgre for example: you would
need the daemon running all the time, possibly interfering with what the
user is doing. Sure you can work around that, but I'm not sure it would
offer any real advantage in terms of speed. You would probably notice
some difference if the repository database was way larger than what it
is and if equo had to run over and over. I think "bigger" db programs
such as Postgre are best suited if you need the extra functionalities
they provide. Sqlite is best fit for these "touch and go" programs
(another example is FireFox).
It would be interesting if you (or somebody else) could provide
profiling data on equo to see how much time it's effectively spending in
the sql queries and if that's really a bottleneck worth spending time
on, but I wouldn't scratch my head too much over it.
mic
On 23/10/2012 06:41, Andre Jaenisch wrote:
Hello, dear list members :)
I had a talk with Enlik and micia three days ago on IRC.
Somehow we came to SQL databases and Enlik pointed me, that Entropy is
using SQLite.
I know, that maybe the mailing list is not the right place for asking
this, but you can read my reasons below.
For now, I would like to know the reasons, why SQLite was preferred
instead of other SQL databases (you know, MySQL, PostgreSQL and so
on).
I know from learning python, that python supports SQLite natively, but
Enlik mentioned some other reasons:
[Saturday, 20. Oktober 2012] [23:43:45 GMT+2]<Enlik>: [...] sqlite is
small, has good support for python, doesn't require any daemon running
(!) - it's just a file on disk - and thus, very safe against cases
when a daemon crashes and Entropy would be in trouble; also, small and
well supported.
However, he is interested in comparing SQLite to other database
management systems concerning to performance.
Why I'm asking?
Well, I thought, it may be useful to documentate the developers'
choices somewhere. Maybe on the wiki. And a mailing list is rather
searchable than IRC. Or is there stored an IRC log somewhere? For
people like me, who are interested in how and why Sabayon works like
it does.
Regards, Ryuno-Ki
