[02:43:43] <dstufft> msabramo: ``pip install readme>=0.5.1 && python setup.py -r -s`` will check your packages long_description to see if it fails rendering
[02:44:25] <msabramo> dstufft: Oh is that new or just something I missed before?
[02:44:39] <msabramo> oh I guess it's new since you said readme>=0.5.1
[02:45:04] <msabramo> will try that out later but thank you in advance; will be very nice to have that
[03:11:32] <dstufft> msabramo: yea it's new. I realized a little bit ago that a setup.py command would be super easy to do and answers some of the questions like "how do I find the long_description"
[20:32:02] <ErikRose> dstufft: Has something changed about PyPI download counts in the last couple weeks? I notice they're 1/10 of what they used to be.
[20:33:07] <carljm> ErikRose: Haven't you heard? Nobody uses Python anymore.
[20:34:29] <ErikRose> sigmavirus24: An interesting idea. You pretty much have to opt into those. Are they catching on that quickly?
[20:35:05] <sigmavirus24> ErikRose: the purported speed benefits are causing a lot of people to try them out. One of the reasons they're faster though is the ability to turn on caching, so maybe?
[20:35:42] <sigmavirus24> It'd be really hard at this point in time to figure out how many people on GitHub are turning it on because I'd basically start collecting data now which is >1 month after the feature was released/announced
[20:36:20] <ErikRose> It would be interesting to get ahold of the logs and see what % a few months ago was from Travis, vs, now.
[20:36:41] <carljm> Hmm, I need to check into that. Last time I committed to one of my OSS projects on Travis, about 1/3 of builds randomly failed with network errors getting the dependencies.
[20:37:56] <ErikRose> That's interesting; I haven't seen that myself.
[20:38:14] <ErikRose> I did a bunch of pyelasticsearch builds yesterday, too.
[20:38:34] <sigmavirus24> I've seen that but it never lasts longer than a day
[20:43:41] <sigmavirus24> carljm: seems to act up every so often
[22:50:16] <dstufft> ErikRose: pip 6.0 caches shit on by default
[22:50:36] <dstufft> I dunno if that'd be 1/10th of what they used to be or not
[22:50:55] <dstufft> but if people's caches are primed now that could be why
[22:51:17] <ErikRose> That strikes me as plausible.
[22:52:21] <dstufft> also pip 6.0.7 which was released on uh, Jan 28th fixed a bug with our on by default cache, previous to that if you used pip in a combination of Python 2, Python 3, and python3.4 you'd get cache evictions
[22:52:49] <ErikRose> Have you seen bandwidth on the CDN drop precipitously?
[22:53:04] <dstufft> (we were using pickle to store the cached items, but we were using the highest pickle protocol available, so pip 3.3 couldn't read a pickle written by Python 3.4, and Python 2.7 couldn't read one written by Python 3 at all
[22:55:58] <dstufft> but it's hard to tell because PyPI's growing in general too
[22:56:45] <ErikRose> Interesting. I forget when I first noticed it. For at least a month or two, my download numbers have been 1/10 of their normal. Maybe my package are all suddenly terrible. :-)
[22:57:29] <sigmavirus24> ErikRose: I haven't checked my download numbers in forever
[22:57:38] <sigmavirus24> You're making me glad I don't =P
[22:57:52] <ErikRose> Yeah, I'll probably replace them with some Google alerts or something.
[22:58:22] <sigmavirus24> holy smokes. alpha release of github3.py 1.0 has nearly 2500 downloads
[22:58:28] <sigmavirus24> that's way more than I was expecting
[22:59:21] <sigmavirus24> of course rfc3986 already has over a .25m downloads
[22:59:21] <ErikRose> sigmavirus24: That's funny; I see 4722.
[22:59:31] <sigmavirus24> ErikRose: I'm using vanity
[23:02:37] <sigmavirus24> Download counts are one way of measuring love, but I get a lot of love from people about github3.py so I generally feel good about spending time there
[23:03:15] <sigmavirus24> Also, the only reason rfc3986 has over 2x the downloads of github3.py in only 8 months is because openstack uses it
[23:03:32] <ErikRose> It's pretty petty; I shouldn't let it bother me so much.
[23:03:34] <dstufft> we generally don't bother because the rolling counts are empheral
[23:03:49] <dstufft> ErikRose: eh, it's a feature of PyPI and if it's not working it should either be fixed or removed
[23:03:54] <ErikRose> But if there was a reason that suddenly really nobody is using my stuff, I'd like to know.
[23:04:06] <sigmavirus24> ErikRose: to each their own. I just feel like if I bothered to put a lot of effort into evangelizing my packages they'd do a lot better.
[23:04:12] <sigmavirus24> Right now I seem to cater to the hipster crow
[23:05:29] <sigmavirus24> I just need to blog about my projects more
[23:05:45] <sigmavirus24> My posts are infrequent enough but when I do blog my projects see a huge boost in stars and downloads I've noticed
[23:06:07] <sigmavirus24> It's just the more I blog the more I want to rant about how no one follows RFCs and it's ruining my life that they don't. yadda yadda yadda
[23:07:09] <ErikRose> I was thinking of registering everythingisterrible.com for my rants and keeping my positive stuff on somesystemsgo.com or grinchcentral.com.
[23:11:24] <sigmavirus24> I need a site that is like "yunofollowrfcs.com" and it's just rageface memes whenever I find a new instance of some developer being lazy and not following a 15 year old RFC
[23:12:13] <ErikRose> I could probably write a whole series of articles titled "I Will Stab ___ In The Face", like http://www.grinchcentral.com/password-rage.
[23:12:30] <sigmavirus24> physical violence is where I draw the line
[23:12:56] <ErikRose> I will stab browser UX in the face, I will stab the cloud in the face...
[23:16:47] <ErikRose> So there's nothing running on the box but what you specifically need, and you benefit from OCaml being non-stupid about bounds checking and such.
[23:16:57] <sigmavirus24> (Looks like C to me: https://github.com/ocaml/ocaml)
[23:17:12] <dstufft> do i have to write things in OCaml if I use MirageOS
[23:21:02] <sigmavirus24> ErikRose: no one uses it?
[23:21:05] <ErikRose> Isn't that just a shell on steroids?
[23:21:15] <ErikRose> sigmavirus24: heh. I wrote a file recovery tool for it in college.
[23:21:39] <dstufft> ErikRose: busybox is a stripped down linux thing yea
[23:22:03] <ErikRose> It does nothing for the kernel or userspace in general; it just merges a bunch of common binaries into the shell, IIUC.
[23:22:22] <ErikRose> openssl, glibc--that all stays.
[23:22:55] <dstufft> ErikRose: yes-ish, but the idea is you don't run all the other crap that tends to run on your typical linux system
[23:23:13] <ErikRose> That's a step in the right direction, then.
[23:23:35] <dstufft> it's still a regular linux kernel and what not, but stripped down so you don't have a whole bunch of random services and crap running
[23:23:46] <dstufft> a lot of people are using busy box docker images to get stripped down dockers