Great, thank you. >> doesn't seem at all something you absolutely NEED redis to do
What alternative do you recommend to think about? Serialize data into session? sqlite? ...? I cannot access the data immediately in controller call (because response of third party service + parsing is too slow). So I need keep lot of data for single session, and give user delayed short info (with links) which data are prepared: - when user requests new page, - with js setInterval() and ajax call Dne pátek 22. dubna 2016 9:23:26 UTC+2 Niphlod napsal(a): > > doesn't seem at all something you absolutely NEED redis to do, but feel > free to use it. > You can use an RConn() instance throughout your entire app. > > On Friday, April 22, 2016 at 8:53:09 AM UTC+2, Mirek Zvolský wrote: >> >> Thank you. >> >> What I need is heavy computing related to few users, they DON'T share any >> data. >> >> User asks a query, some web service gives very long response, >> and it is necessary to parse it maybe 20s. >> In this time user can continue work include send next questions. >> >> Finally user will receive info, which answers are parsed allready >> and then he can choose from parsed data. >> But I don't want save all parsed data to postgres database. >> >> I think redis could be good here? >> With namespaced keys instead of separated db, as you recommend. >> >> So I need know, what is minimum for such scenario. >> Redis + python module installed of course, that already works nice from >> python. >> >> But in Web2py? Is it necessary to think about threads/queues ? >> Or import the redis module and make one instance as bellow is enough: >> >>> import redis >>> r = redis.StrictRedis() >>> r.set(....) >> >> >> >> >> >> >> >> >> Dne čtvrtek 21. dubna 2016 22:27:28 UTC+2 Niphlod napsal(a): >>> >>> that machinery is only needed for a module that needs to be used within >>> web2py, as cache and session need (and should) use a single object per >>> class with namespaced keys. Also, there's really no point or any >>> performance benefit on using separate databases in redis, and they're >>> unofficially deprecated because they're not working reallty fine with the >>> new redis cluster. it's far better to namespace your keys. >>> >>> tl;dr: no, RConn() wasn't made with databases preselection in mind, >>> mostly because there's no need to. >>> >>> On Thursday, April 21, 2016 at 7:41:43 PM UTC+2, Mirek Zvolský wrote: >>>> >>>> I try understand how redis support should work. >>>> >>>> In the contrib web2py code I see function RConn() with code inside: >>>> >>>> >>>>> instance_name = 'redis_conn_' + current.request.application >>>> >>>> if not hasattr(RConn, instance_name): >>>> >>>> setattr(RConn, instance_name, StrictRedis(*args, **vars)) >>>> >>>> >>>> It looks like this should separate (isolate) redis/StrictRedis >>>> instances by request.application. >>>> >>>> However, if I create 2 redis/StrictRedis instances and save them as 2 >>>> different variables, >>>> they will share same data. >>>> To really separate them, I need set different db=NNN parameter: >>>> >>>> r1 = StrictRedis() >>>> r2 = StrictRedis() >>>> will be same, >>>> >>>> but >>>> r1 = StrictRedis(db=1) >>>> r2 = StrictRedis(db=2) >>>> will be different. >>>> >>>> So, is the per application isolation really working? >>>> >>>> -- Resources: - http://web2py.com - http://web2py.com/book (Documentation) - http://github.com/web2py/web2py (Source code) - https://code.google.com/p/web2py/issues/list (Report Issues) --- You received this message because you are subscribed to the Google Groups "web2py-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to web2py+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
