[Tilesathome] Hourly job at minute 25

[Dirk-Lüder Kreie]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Norbert Hoffmann schrieb:
> Dirk-Lüder Kreie wrote:
> 
>> Norbert Hoffmann schrieb:
>>> Christopher Schmidt wrote:
>>>
>>>> This is the requeueer, which requeues tiles which have been out for more
>>>> than 2/6/24 hours.
>>> Ah, thank you. I would suggest, that those tiles would be requeued at the
>>> top of the queue (or at a higher priority) - so they wouldn't have to wait
>>> for another 3 weeks until rendered.
>> I strongly oppose that.
>> There is almost always a reason why these tiles failed. and any request
>> (manual or automated from the changedTiles daemon) will update the prio
>> to nicer values anyway.
> 
> It is *one* new try after every 24 hours I suggest. (I am only talking
> about tiles, that have been requested by the requeueer). So you are right,
> that "higher priority" may not be the way to go. But I think "top of the
> queue" after those timeouts is sensible.
> 
> If you fear, that error tiles get rendered again and again, then the
> problem are more those requests at higher priorities where the client
> *reports* errors immediately. Perhaps *those* should not be requeued at
> once (as it is now) but too be taken care of by the requeueer.

I think one of the main problems of the t at h project is that the meta
information is too fragmented. I do know most stuff about the client,
but only so much about what is going on serverside.

The client has the definite capability of reporting errors other than
simply re-requesting the offending tile, just that there is no
serverside part (that I know of) that would make use of it.

Another problem is, that the development of server and client were a
long time only in "firefighting mode" so to speak. Always only fixing
the most broken stuff, without really adapting how the entire thing works.

There are also a lot of ideas how to do things better, but no time (or
skill) to actually code what's necessary.
For example we would need a fuction to manipulate OSM data in such a way
that ways starting and ending outside of a bbox that are not areas are
discarded along with untagged nodes that are not part of any way
crossing/touching the bbox.

- --

Dirk-Lüder "Deelkar" Kreie
Bremen - 53.0952°N 8.8652°E

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFH/qVHFUbODdpRVDwRAhQ6AKCLjy7U2r1JZcve1gMErO2GyqLkBQCgwcv7
bue41bLDTEMPDY2AGFaN9yc=
=jm0x
-----END PGP SIGNATURE-----

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3743 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://lists.openstreetmap.org/pipermail/tilesathome/attachments/20080411/12d2572a/attachment.bin>