Some workunits have been loaded. These are testing.

I asked a question about uploading of aborted results.

Some workunits have been loaded. These are testing.

I am glad that the testing has begun.

The application seems to work. Already found 2 ODLK:
DRwmjE6YqSYbZGpCVCf22jzvn # rule=25
4K8iYyrbV6vvb5bpRJnQAJ7LY # rule=07
HNKv5Chy8LiU4eJTV7pk6PLL2 # rule=32

The workunits should be 1 hour long approximately.
Checkpoints are implemented: if you shut down computer, progress will be saved and then resume again.
Application also checkpoints when aborted, but it does not work yet (see 3309).
Progress reporting is not implemented: but boinc can estimate progress well enough.
I am very confident that there are no errors, but testing is needed.

More workunits will be generated when I am ready.

The application seems to work. Already found 2 ODLK:
DRwmjE6YqSYbZGpCVCf22jzvn # rule=25
4K8iYyrbV6vvb5bpRJnQAJ7LY # rule=07
HNKv5Chy8LiU4eJTV7pk6PLL2 # rule=32

Already three ODLK? :)

Unique solutions!

Started working on the assimilator program (to copy results to database and grant credit).

I generated more of the workunits. The application is marked beta - that is, only crunchers that have beta enabled will get the workunits. This is intentional. If you (cruncher) do not want to run the beta application, turn off beta in your preferences. The validator/assimilator is not finished - therefore no credit will be assigned yet. Again if you (cruncher) do not want to compute workunits that do not yet have credit, turn off beta.

I will try to work in this order:
* finish validator program
* compile application for windows
* write announcement that project is running
* automatic publishing of results

The source code is available: in search5 branch, search5 folder of my repository. Everyone is invited to help. Or you can look at the code and spot bugs (in search5.cpp program).

Validator is working and credit is assigned. Please check your tasks.
Found 1051 ODLS.

I sent task tot5_51a_T72KJVr2ZHr1b8gfi6roQQfmq_1 to test how the validator behaves when unknown (or duplicate) result is received.
- it should not crash, it should still save the results, but not update the segment.

I updated the server status page to show more info about batch tasks: not only unsent and in-progress tasks, but also:
tasks ready to validate, invalid tasks and crashed tasks.

Before this project is launched, it is essential to coordinate efforts with the other projects in DLK10 space.
Which ruler to search? Currently, rule 51 is used for testing, but it can be easily changed.
Which sub-space to search? If there are other projects running, we should choose non-overleaping search space. This is not so straightforward.
I will ask Natalia in Progger's forum, because she has been experimenting with few other rulers in manual mode with XAVER and Demis.

Minor issue: the validator is not updating the host-app-version table on the server.
Ok, now it is updating, but still not fully.

5381 ODLS found. 5945977342367 SN DLK, 819657164 KF DLK.
Checking for uniqueness and post processing (zamyk) is not done yet.

Finish file present too long: 148278

Work generation suspended, except two segments. Great work!

5381 ODLS found. 5945977342367 SN DLK, 819657164 KF DLK.
Checking for uniqueness and post processing (zamyk) is not done yet.

These are great results!
It is interesting to see the processed results (zamyk).

I tested the experiment PADLS TOTAL rule 51 on my PC.
Approximately 15 millions CFs checked, 240 unique CFs ODLS found.

Tomáš Brada
I posted these results so you can compare your results with mine.

Link
https://cloud.mail.ru/public/3Bnv/5FYgynNeC

I exported the odlk database. But the format is very bad. In automatic publishing of results, the format will be better.
https://boinc.tbrada.eu/download/tot_odlk_1.sql

I deleted tasks of old experiments from the database to save space and improve speed.