@fabien Oh I am very pleased to see you again in the forum as there were some feeling of one way communication throughout the forum for some time (asking questions, but not getting answers) with a bit more complicated questions. :)
We are working on 4.2 branch for our developments, and probably will switch the systems to 5.x when migration script will be available. Well I understand that one needs to make money, but selling basic things actually is harmful to the whole project, just think - there will always be setups with 5M of customers, which will require fine tuning, probably this is the right field to raise money. Again - no hard feelings... About multithreading, as you can see from my posts on this thread, it seems to be that TERP is like single threaded app. Because simple scheduled task which update data models (taking data from external data source), block the whole TERP process, no one is being served till the function ends (?!?). Still we noticed that the same scheduled functions sometimes run in parallel creating concurrency. And what happen if several Application servers are run against single DB, there is no locking - again scheduler will fire up event on every server... This is noticed things, have not looked into code, as have been busy with other things. I have no doubt that 5.x are a lot better than 4.2 :), but we have several 4.2. setups and some major localization developments which are not rational to move them, as some part of them probably will not be required anymore. About DB improvements - I made some indexes on it and got more or less impressive results (we have no very big setups for the time being, to test with). It is a pity that there is no way to place them in a module - no initialization part when module installs, or come other way to place them on the DB. It would be interesting to hear why TERP is being developed exclusively against single DB engine, as there is actually no specific postgres functionality utilised. I do not say that Postgres is good or bad, but at least MySQL do not require so much attention to sustain performance. Postgres without attention just eats himself up in a day or even few hours on a small scale setup, which is not the case with MySQL... At least I have not noticed similar in my almost 10 year experience with it. sraps -------------------- m2f -------------------- -- http://www.openobject.com/forum/viewtopic.php?p=34396#34396 -------------------- m2f -------------------- _______________________________________________ Tinyerp-users mailing list http://tiny.be/mailman/listinfo/tinyerp-users
