Revision history [back]
Hi there, this will take a while (at least if I start working on it - I've never touched app engine yet).
The main issue is adaptation of code from using Django ORM to GAE datastore. There is something called "django-nonrel" adapter that is in development out there. This may help. Last time I read about it "nonrel" could not "translate" all django orm queries to datastore but could handle the most frequently used queries.
I am gradually removing ORM calls from the application code so that eventually one might re-implement the model layer to work on the engine. IMO the best way to proceed it to first isolate application code (the view functions, templates etc.) from the direct model lookups, rather than just rewrite everything specifically for GAE (because in this case you will end up with two incompatible applications).
If you are interested to run a site - you will not suffer in any way by using old school SQL-based database for quite a while. With careful caching in the application you can sustain very high traffic (~1000-3000 views per second and dozens of posts per second). Most forums, including the best ones are much slower.
non-technical consideration:
App engine is free to the entry level users, which is maybe part of the appeal. Traditional hosting of decent quality costs ~ $10 a month (extra cheap offerings usually bring more pain than savings). sure $10 a month will eventually add up, however that expense will be minor compared to the effort maintaining the forum.
The bottom line is that eventually the app will be portable to GAE, but after other important issues are solved and the application is prepared to porting to other data backends.
Hi there, this will take a while (at least if I start working on it - I've never touched app engine yet).
The main issue is adaptation of code from using Django ORM to GAE datastore. There is something called "django-nonrel" adapter that is in development out there. This may help. Last time I read about it "nonrel" could not "translate" all django orm queries to datastore but could handle the most frequently used queries.
I am gradually removing ORM calls from the application code so that eventually one might re-implement the model layer to work on the engine. IMO the best way to proceed it to first isolate application code (the view functions, templates etc.) from the direct model lookups, rather than just rewrite everything specifically for GAE (because in this case you will end up with two incompatible applications).
If you are interested to run a site - you will not suffer in any way by using old school SQL-based database for quite a while. With careful caching in the application you can sustain very high traffic (~1000-3000 (~3000 views per second and dozens of posts per second). Most forums, including the best ones are much slower.
non-technical consideration:
App engine is free to the entry level users, which is maybe part of the appeal. Traditional hosting of decent quality costs ~ $10 a month (extra cheap offerings usually bring more pain than savings). sure $10 a month will eventually add up, however that expense will be minor compared to the effort maintaining the forum.
The bottom line is that eventually the app will be portable to GAE, but after other important issues are solved and the application is prepared to porting to other data backends.
Hi there, this will take a while (at least if I start working on it - I've never touched app engine yet).
The main issue is adaptation of code from using Django ORM to GAE datastore. There is something called "django-nonrel" adapter that is in development out there. This may help. Last time I read about it "nonrel" could not "translate" all django orm queries to datastore but could handle the most frequently used queries.
I am gradually removing ORM calls from the application code so that eventually one might re-implement the model layer to work on the engine. IMO the best way to proceed it to first isolate application code (the view functions, maybe templates etc.) from the direct model lookups, rather than just rewrite everything specifically for GAE (because in this case you will end up with two incompatible applications).
If you are interested to run a site - you will not suffer in any way by using old school SQL-based database for quite a while. With careful caching in the application you can sustain very high traffic (~3000 views per second and dozens of posts per second). Most forums, including the best ones are much slower.
non-technical consideration:
App engine is free to the entry level users, which is maybe part of the appeal. Traditional hosting of decent quality costs ~ $10 a month (extra cheap offerings usually bring more pain than savings). sure $10 a month will eventually add up, however that expense will be minor compared to the effort maintaining the forum.
The bottom line is that eventually the app will be portable to GAE, but after other important issues are solved and the application is prepared to porting to other data backends.