On 11/01/2015 09:43 AM, Chris Angelico wrote: > On Mon, Nov 2, 2015 at 3:24 AM, rurpy--- via Python-list > <python-list@python.org> wrote: >> I dont recall seeing anyone posting asking why they could not get >> Python to install on Windows 95 recently. I only read this group >> intermittently but I have seen *many* posts asking why they couldnt >> install on XP. >> >> You acknowledge yourself: "there are a lot more XP boxes out there." > > Yes, there are. This is not a customer base; they are not paying me, > nor (as far as I know) the PSF, to support them.
I wasn't asking *you* to do anything. And I wasn't asking for python to support XP. I'm not sure how you have possibly got that idea. In fact I wasn't *asking* anyone to do anything. I was pointing out (the obvious) that there was a problem with people not understanding that Python-3.5 will not run on XP and suggesting a low-cost way to reduce that problem. >>> There *is* a plan to have the installer give a better error message >>> for this situation. >> >> A better message from the installer is necessary but not sufficient. >> Don't make people go through the effort to download the whole thing, >> do their planning and preparations for using or upgrading Python >> only to discover at the last moment it wont work. > > The largest installer [1] is 30MB, which might have been considered > large a decade ago, but on today's connections, that's probably going > to download in less time than it takes people to search for "python", > find python.org, and decide to click the download link. The web > installer is less than one megabyte. If you can't afford to download a > single meg of executable to find out whether something works on your > system, you probably can't afford to download the page with the info > telling you not to bother downloading the binary. As I said it is not solely the number of minutes needed to download something. And not everyone in the world has high-speed reliable connection. It wasn't that long ago my internet connection was a modem. And even today with screaming fast 100KB/s (sometimes) connection I often have to babysit a download to get it to complete or deal with long outages. There are plenty of places much less connected than I am. And of course those are the same places where people are more likely to still be using XP. >> That is really shitty customer relations. > > See above, and define 'customer'. customer: the people to whom you are providing a product or service. Of course, unless the PSF has a charter that imposes some legal requirements, PSF/Python development community has no legal obligation to do anything for anyone. They could add spam-ware to the installer, make backward incompatible changes on a whim, release software that crashes when started. Now could we return to reality please? >> The reality is that people trying install Python-3.5 on XP *is* a >> problem. Telling them they should have read some obscure release >> notes is not a solution. > > Nor is telling them they should have read the web site that they > downloaded it from. Remember, people can click a direct download link > *on the python.org front page* and be immediately downloading Python 2 > or 3 for the OS that the browser announces. Where is that? I didn't see anything that wasn't accompanied with text that could accommodate a windows xp warning. But then I was looking at it from a Linux machine. And even if it is impossibly awkward to provide in one case (though I doubt that's the case), there certainly should be a warning on the windows specific pages. AFAIK, the only place where dropping XP support is mentioned in the docs is way down the bottom of What's New. PEP-11 doesn't count -- it unrealistic in the extreme to expect a Windows user who wants to try Python to read (or even know about) the PEPs, or that one of them describes supported OSes. > Where would you put the > big fat noisy warning? I would not put a "big fat noisy warning" anywhere. I would put a minimally sized but clear notice where the windows version of python-3.5 can be downloaded. >>> But I don't think the web site necessarily has to >>> have noise about old versions of OSes. Where would you draw the line? What noise? Providing basic useful information to a significant number of customers/clients/users, information that is not provided elsewhere in the normal course of their actions, is not "noise". If you want to get rid of noise, how about ditching corporate-style PR like "Success Stories", or a big graphic of fibonacci numbers in python? It's really sad to see open source projects taken over by marketing wanna-bees who think PR fluff is more important that actual actionable information. >> I think my responses above answer that. > > Not really, no. There are currently a large number of XP boxes out > there, for some definition of 'large'. Presumably that number is > dropping. At what point will it be appropriate to ditch the warning? > And what about older versions of non-Windows OSes - if there are a > large number of people still running an old Mac OS, should we include > something on the front page that warns everyone about a lack of > support? Why don't you try answering that question yourself? Start by asking how many posts to this list asking why the poster couldn't install Python-3.5 on an old MacOS system you've seen. > How large is large? I say again: Where would you draw the > line? If you insist on a concrete number (I presume so you can pick nits with it rather than the general proposition), 5 years. Then reevaluate. Once someone has actually adds a note dispute about whether it should be there will fade into nothingness and removing it after xp truly is insignificant will be contentious. -- https://mail.python.org/mailman/listinfo/python-list
