So there's 3 choices here: a) revert the importwarning entirely b) make it suppressed by default c) more complicated code in import.c to only emit the warning if the import fails. After a bit of a chat with Neal, I think the best combination of prudence and functionality is (b). (a) also works for me. (c) is a bit too scary, post-beta. import.c is a bad place where programmers go to die - I'd rather not mess about with it at this stage of the release cycle. Unless I hear screams in the near future, (b) is what's going to be done for beta2, and therefore 2.5 final. We can re-evaluate the situation for 2.6 - maybe a more complex solution like (c) can be done for that. This means Google can just turn it on in sitecustomize.py and Guido can avoid the hordes of peasants with pitchforks and burning torches. Anthony -- Anthony Baxter <anthony@interlink.com.au> It's never too late to have a happy childhood.
At 01:24 PM 7/6/2006 +1000, Anthony Baxter wrote:
This means Google can just turn it on in sitecustomize.py and Guido can avoid the hordes of peasants with pitchforks and burning torches.
Is that really true? It seems to me that Guido indicated a sitecustomize-solution wasn't possible, in which case suppression is no better than elimination. By the way, while I haven't formally reviewed any of the patches to delay the warning, the one I saw that went by on Python-Dev looked quite straightforward and minimal. Until I saw it, I also believed that trying to implement something like that would be difficult and delicate. But the patch I saw looked very simple and direct.
On 7/6/06, Phillip J. Eby <pje@telecommunity.com> wrote:
At 01:24 PM 7/6/2006 +1000, Anthony Baxter wrote:
This means Google can just turn it on in sitecustomize.py and Guido can avoid the hordes of peasants with pitchforks and burning torches.
Is that really true? It seems to me that Guido indicated a sitecustomize-solution wasn't possible, in which case suppression is no better than elimination.
Don't worry about it. It's irrelevant anyway since we're not upgrading to 2.5 any time soon. BTW The pitchforks pitch was widely misunderstood -- it was a dramatization to emphasize what I still consider a general problem; if I believed the problem was limited to Google I would have proposed an internal solution. -- --Guido van Rossum (home page: http://www.python.org/~guido/)
participants (3)
-
Anthony Baxter
-
Guido van Rossum
-
Phillip J. Eby