Chris Withers wrote:
I think documenting that something is going to go away is useful, but ultimately, people only really worry about it when something stops working.
I think it's not that simple. Look at the amount of work people put into the Zope stack in recent month to get rid of DeprecationWarnings for the sha, md5 or sets modules or various other warnings emitted by Python 2.6. There's lots of energy spent on those warnings *before* they break anything.
I've got way to bored to the millions of meaningless deprecation warnings emitted by *released* zope/plone/blah versions over the years...
Yes, me too. But I feel our use of the system wasn't very good. I have seen people putting in deprecation warnings for things they didn't like, but without there being an alternative. At other times we deprecated particular single methods of something to move them some place else without much clear win. To me it seems that we get warnings from the Python level on a somewhat broader level, like: the entire package X is going away, use Y now. The message itself is short and concise, but you can look up detailed documentation about the new package and its API. There must be something to learn from that. Hanno