Then they better get some hardcore Django people onboard. The reason (well, one of the reasons) that Heroku was so successful was that the top people were Ruby hackers themselves.
Unlikely. Development speed is an inverse function of company size after a certain point. Startups will eat Salesforce's lunch at developing something new.
GAE looks like this at first glance. But it turns out to be focused 90% on scalability, and 10% on easy deployment. The main issue for Django developers is that on GAE, you're using a nonrelational database. Assumptions about the RDBMS run deep in Django and in most Django apps, and great effort (see django-nonrel) is required to present even the thinnest veneer of compatibility.
In my experience (>2 years of GAE consulting for paying clients) I think using GAE is probably the wrong choice for most new projects. If you're new, you need to be able to iterate ideas quickly and you will value tools that (1) give you flexibility, (2) offer a lot of useful reusable libraries, and (3) have a large and stable culture (good docs, stable APIs, people you can hire). And if you have a scalability challenge somewhere down the road, it would make perfect sense to rewrite most of your app on a platform like GAE -- after all, by that point you'll have the basics of design quite solid, so you don't need as much flexibility, and every other part of the effort gets easier when you have money to throw at it.
It is a common misconception, though, that AppEngine serves the same needs as Heroku, or is a Heroku equivalent for Django. I think that's the main reason we didn't see more Herokus-for-Django* earlier.
They give you free-ness up to a rather large number of pageviews/users/whatever, right? Could the process be smoothed to make the workflow as smooth as Heroku makes it, or do the differences in Google's infrastructure make that too much of a pain to automate?
Except that Heroku for Django in the long term is likely to be Heroku (given they've said in the past they plan to expand to other languages).
Then they better get some hardcore Django people onboard. The reason (well, one of the reasons) that Heroku was so successful was that the top people were Ruby hackers themselves.
Unlikely. Development speed is an inverse function of company size after a certain point. Startups will eat Salesforce's lunch at developing something new.
Then Salesforce can just buy them.
Djangy has 600+ registered beta users, and we're inviting more all the time. Check us out!
http://www.djangy.com
Sign up to try it out... http://djangozoom.com :)
I thought that was Google App Engine?
GAE looks like this at first glance. But it turns out to be focused 90% on scalability, and 10% on easy deployment. The main issue for Django developers is that on GAE, you're using a nonrelational database. Assumptions about the RDBMS run deep in Django and in most Django apps, and great effort (see django-nonrel) is required to present even the thinnest veneer of compatibility.
In my experience (>2 years of GAE consulting for paying clients) I think using GAE is probably the wrong choice for most new projects. If you're new, you need to be able to iterate ideas quickly and you will value tools that (1) give you flexibility, (2) offer a lot of useful reusable libraries, and (3) have a large and stable culture (good docs, stable APIs, people you can hire). And if you have a scalability challenge somewhere down the road, it would make perfect sense to rewrite most of your app on a platform like GAE -- after all, by that point you'll have the basics of design quite solid, so you don't need as much flexibility, and every other part of the effort gets easier when you have money to throw at it.
It is a common misconception, though, that AppEngine serves the same needs as Heroku, or is a Heroku equivalent for Django. I think that's the main reason we didn't see more Herokus-for-Django* earlier.
*Disclaimer: I am a founder of http://djangozoom.com/ .
http://djangy.com is one
Google App Engine is similar, but not identical to normal Django.
They give you free-ness up to a rather large number of pageviews/users/whatever, right? Could the process be smoothed to make the workflow as smooth as Heroku makes it, or do the differences in Google's infrastructure make that too much of a pain to automate?
2 replies →