> The biggest issue I have is that model syncing fails
> silently (it's as if the app doesn't exist) if an
> import in the app models file dies with an exception
> (for instance, not found).
I've wasted hours trying to figure out those kinds of silent failures... it's not much fun. Another one that stopped me in my tracks for about an hour is that an app must have a models.py file to run unit tests, even if your application doesn't have any models. If it's not there, instead of just saying 'models.py not found' it says something like 'application <your app name> could not be found'. As others have said though, I guess that's the price you pay for the luxury of having a framework to do the low-level dirty work for you.
I don't consider the tests.py issue to be "part of using a framework" its just as bug. However solving the problem properly is a harder problem (right ow djangos app cache is a part of the orm, the right fix is to have a real app cache that doesn't kow about models, ad let the orm have its ow cache that just deals with models).
but it doesn't do a good job of explaining what to do when your database isn't up to date. Geo django also gives horrible error statements (none at all when the libraries aren't installed right).
I'm the author (although not the submitter), I didn't intend this as any sort of flame-bait, I was just curious how well Django did at this stuff (a long time ago I remember seeing a better error message wiki page).
Great article! I wouldn't worry about the people who didn't read the article and are getting upset about the perceived content, maybe they will learn an important lesson about jumping to conclusions. While it's obvious that the title is what is causing the negative comments, having actually read the article I have to say that the title is appropriate. Thanks for writing this up!
No, I read it and thought it was much more reasonable than the Rails one with a similar title. The problem was that it was in response to the Rails article, which was of very low quality.
No offense but I'm getting the impression that you (and grandalf) haven't read past the title here.
Nobody is saying "[Django|Rails] fails as a tool/project/framework", the topic here (and in the related submission about Rails) is how do the tools we use respond to error, how clear is the message we are given and weather or not these responses could be clearer or friendlier. In short, "useful error messages".
I think it's an interesting and useful topic of discussion and I submitted this link to be complementary to the story about Rails (though I admit I probably should have just added it in a comment to that story), not to some how put Django against Rails (which seems to be a point of misunderstanding) or to put Django itself down (which wasn't anyone's intent).
To be fair, he could ask if the article is worth reading because he felt turned off by the title. The title is at least a bit inflammatory and the article is the complete opposite of it.
> The biggest issue I have is that model syncing fails
> silently (it's as if the app doesn't exist) if an
> import in the app models file dies with an exception
> (for instance, not found).
I've wasted hours trying to figure out those kinds of silent failures... it's not much fun. Another one that stopped me in my tracks for about an hour is that an app must have a models.py file to run unit tests, even if your application doesn't have any models. If it's not there, instead of just saying 'models.py not found' it says something like 'application <your app name> could not be found'. As others have said though, I guess that's the price you pay for the luxury of having a framework to do the low-level dirty work for you.