alex23, 13.01.2012 06:41: > On Jan 13, 3:02 pm, Steven D'Aprano wrote: >> Why is it that only Linux and Mac users are accused of being "zealots"? > > Oh please. Don't tar me with the Windows brush. I'd have used the same > term no matter what OS was being recommended. > >> If I ask how to install (say) MYOB or Photoshop on Linux, and people tell >> me that I will have to use Windows if I want to easily run that software, >> I don't accuse them of being a zealot. > > And if lxml didn't have Windows binaries, then maybe this would be the > case. [...] > In what way is downloading pre-built binaries and then installing lxml > on Windows like driving across the ocean?
Note the two sides to this problem. On the one side, there is a Windows user who is unable to easy_install lxml because the latest binaries are not on PyPI. It's an obvious first reaction to blame the project for not putting them there. On the other side, there's the lxml project which can't currently produce binaries due to the lack of a proper maintainer who has a legal and properly configured copy of that commercial operating system installed. It's an obvious first reaction to blame the Windows user who is unable to configure his/her own system to properly build lxml there. Who's to blame here? Personally, I think that solving the problem on the second side is much closer to actually driving across the ocean, whereas the first side sounds more like it's recommending it. But I guess that's just because I *am* on the second side. Now, you are saying that this problem is easily solved by downloading binary packages from an unknown third-party source and installing them on your machine. I've already seen requests to upload those specific binaries to PyPI a couple of times. This appears to be a normal way of thinking specifically amongst many Windows users, simply because this system trains them to download opaque binary software packages from arbitrary external sources in order to install them manually and use them. Coming from a Linux angle, where signed and trusted software has been the normal case for more than a decade now, so much that an attempt to install unsigned or untrusted software actually leads to a clear warning, I cannot see this as a reasonable recommendation. It may well be that Christoph Gohlke is generally trustworthy enough to install the software he builds in a given user setting. However, presenting this as the ultimate solution to any software installation problems is clear zealotry in my eyes, because it ignores the fact that software from unknown third-party sources may simply not be a valid alternative in a given setting, e.g. on a company computer. As usual, it's not all that easy, but IMHO, recommending to use Linux isn't that much worse than recommending to install untrusted binary software. Stefan -- http://mail.python.org/mailman/listinfo/python-list
