deprecate Spider.make_requests_from_url

[kmike]

Hey,

Spider.make_requests_from_url is a shortcut for a single use case: sending initial requests from a spider with start_urls attribute. It saves a single line of code (for url in start_urls) in one specific use case, but yet we have to document it, users have to read about it and understand what it is for.

The name suggests it is more general - but we can't yield Link objects and get requests created by make_requests_from_url, it does nothing for CrawlSpider (I'd expect URLs to be passed through this method), and it is ignored for SitemapSpider. This is inconsistent.

What about deprecating make_requests_from_url and removing it from docs? IMHO it will make API simpler, this hook does nothing useful now, and if we ever want a general 'process URL and get a Request' method we'd have to use another name for backwards compatibility anyways.

[nramirezuy]

I agree; this method isn't that useful. Also now you even have a shorter import for Request if you wanna be lazy (:

[chekunkov]

+1