Hi all,
so I switched all read only tests on trunk to "one time setup". This
made it so that the time required for building geoserver on my machine
halved, going down to 3 minutes 30 seconds (I have the wfsv online tests
running that do take 30 seconds and that most of you don't have enabled
I guess).
Now, I'm going to take some time wrinting two lines of docs in the wiki.
Can you test the modifications as well (once Simone fixes the build,
that's now broken due to a missing method)?
Also, given that this big commit does not require any java5 support...
shall we switch 1.6.x as well?
Cheers
Andrea
Andrea Aime ha scritto:
Hi all,
so I switched all read only tests on trunk to "one time setup". This
made it so that the time required for building geoserver on my machine
halved, going down to 3 minutes 30 seconds (I have the wfsv online tests
running that do take 30 seconds and that most of you don't have enabled
I guess).
Hudson seems to appreciated as well... build time went down from "5 minutes 34 seconds" to "2 minutes 48 seconds". Not bad! 
Cheers
Andrea
Andrea Aime ha scritto:
Hi all,
so I switched all read only tests on trunk to "one time setup". This
made it so that the time required for building geoserver on my machine
halved, going down to 3 minutes 30 seconds (I have the wfsv online tests
running that do take 30 seconds and that most of you don't have enabled
I guess).
I just added some documentation on the current testing framework here:
http://geoserver.org/display/GEOSDOC/1+Unit+and+functional+tests
It's not very detailed, but I guess better than nothing. Reviews
and additions to the page welcomed
Cheers
Andrea
Andrea Aime ha scritto:
Hi all,
so I switched all read only tests on trunk to "one time setup". This
made it so that the time required for building geoserver on my machine
halved, going down to 3 minutes 30 seconds (I have the wfsv online tests
running that do take 30 seconds and that most of you don't have enabled
I guess).
I switched 1.6.x too... unfortunately the effect it not as nearly as
visible as on trunk... wondering why, I checked, the one time
setup is working as expected... maybe on trunk the xml schema
cache lives for as long as the jvm that's making the tests, whilst
on 1.6.x it just lives as long as the test lives... but I'm
not sure (parsing the schemas the first time is _very_ expensive).
Cheers
Andrea