Hello Timur,
I am sorry. I was so focused on Tracker that I did pick a wrong subject.
The goal is to develop an open /community/ _concept_, how to handle data
on the phone.
It is not picking a tool and try to get it up and running.
I see tracker as a efficient tool to handle Metadata. Metadata to point
you in the right direction.
We could use tracker to have a easy way to categorize Data, independant
where they are stored.
Some information can still be stored in SQL. (Does not make sense for
Contacts, but maybe for the message body of an email?)
Lucene seems for me more a piece of software that is focused on
searches. I want to avoid searches on the phone as much as possible.
Believe me as an expert in chaos I know how nasty a search can be.
Mongo DB is to big, not focused on embedded system. If I have to write a
similar system I would maybe port Bitsy as a base point. I like KISS
aproaches.
However with tracker I have a ready to use indexing concept I can start
with. It is beyond what I had originally in mind. Since it moonshots my
Idea,
it is worth to give it a chance. However I am very sceptical on the
Gnome people ("implementor") and the KDE people ("Concept designers")
when they come together in the same tool?
Well that is interesting an unpredicted development.
One last sidenote on google and moonshots: Even if I do not belive in
Googles goals, I think they do have the right concept to reach their goals.
Something worth to copy!
Keep challanging! (lol)
Peter
On 06.06.2016 00:42, Timur Kristóf wrote:
Hey Peter,
This topic has been on my mind for some time, I just haven't had the
time to bring it up yet. I'm definitely interested.
In the past I spent a lot of time working with a CMS that used a non-
relational content repository. My experience tells me that for most use
cases, you don't actually need an RDBMS; a NoSQL solution could get the
job done in a simpler-to-use and more efficient way.
However, as yet I'm unaware of a suitable piece of software.
Here are the most notable ones I've looked into:
* MongoDB: could be an obvious candidate, but it was designed and
optimized for very different use cases, so not something I'd run on my
phone.
* Tracker: looks like a bloated monster, and under the hood it too just
uses sqlite. So it definitely isn't more effective than just using
sqlite as is. (Side note: on Harmattan, Nokia chose to store literally
everything in Tracker... which wasn't such a good idea.)
* Lucene: unfortunately the main implementation is in Java, and the
native version (clucene / lucy) is lagging behind hopelessly. (Side
note: IIRC, the Qt project uses clucene, maybe just because they didn't
find a better alternative.)
So, the ultimate solution still eludes us.
I believe a "lucene-like" approach might be right way to go, but at
least I haven't yet found a decent implementation that I would be
comfortable going with. Maybe we would have to "roll our own"
eventually.
Best regards,
Timur
On v, 2016-06-05 at 23:39 +0200, Peter Kovacs wrote:
I changed the topic to something more apropriate.
I took a brief look at tracker, and the goals and alround project
sounds
interesting. It is a slight different direction I had in mind,
thought.
But well it reminds me on how apple should have taggled their iOS
Datamanegment but failed to do this.
@Chris Adams: Do you refer on Problems with tracker within Projects
on
Desktop or Embedded environments?
I think if you get Issues with DBus (seems to be core feature for
Tracker, yay) you get lost quickly without in depth knowledge. Maybe
you
refer to that?
However I think if I try to make tracker ready for Sailfish as an
alternate way for storing Data instead of SQL RDBMS, you need more
then
just tracker.
A Data API that connects QT / QML - way to handle Data with tracker
is
vital for such a Project. And that in a way you can use the API with
80%
plus of the use cases when it comes down to store Information on your
phone.
The API should make this tool work seemlessly. And I agree with
Chris
Adams, right away, that this is realy challenging if this should
beat
the experienced SQL way.
I have some Ideas thought. And at least a broader vision on what I
had
before.
I will write some sort of concept in the upcomming weeks.
Is besides Andrej anyone else interested to share Ideas, requirement
he
sees in respect of MetaData/ Data storage in general?
If it is only a small circle I would discuss this off the list, and
come
back to you all if the concepts is ready. Then I would see a PoC and
at
latest a Warehouse version.
currently this is based on hobby/freetime level. so I will need some
time.
Thanks for the input so far!
_______________________________________________
SailfishOS.org Devel mailing list
To unsubscribe, please send a mail to devel-unsubscribe@lists.sailfis
hos.org
_______________________________________________
SailfishOS.org Devel mailing list
To unsubscribe, please send a mail to devel-unsubscr...@lists.sailfishos.org
_______________________________________________
SailfishOS.org Devel mailing list
To unsubscribe, please send a mail to devel-unsubscr...@lists.sailfishos.org
