Re: Performance with and without the garbage collector
Steven D'Aprano writes: > ... > Is anyone able to demonstrate a replicable performance impact due to garbage > collection? I have had some experience with the performance impacts of garbage collection -- not completely replicable but rather frequently visible. Huge Zope instance (with around 2 GB heap usage; millions of objects); a badly designed (external) component that created thousands of temporary objects during searches. What we could observe: occasionally, our Zope got unresponsive for about 1 minute; a monitoring software typically told us that at least on of Zope's worker threads was in the external component mentioned above. The analysis finally revealed: creating and releasing thousands of short lived temporary objects frequently called for a garbage collection. The garbage collection prevented all other Python work while running -- and this caused outages in the order of a minute. -- https://mail.python.org/mailman/listinfo/python-list
Re: Wanted Python programmer to join team
On Tuesday 17 May 2016 16:18, Marko Rauhamaa wrote: > Steven D'Aprano : >> Personally, I think that advertising a job position without saying who >> you are, what you do, and offering at least an indicative salary >> range, are *astonishingly* rude > > I don't believe they care. > >> (to say nothing of counter-productive). > > Maybe, maybe not. > > > I bet the zebras on the savannah consider the lions astonishingly rude > and their strategy counter-productive. The savannah would be a nicer > place if the lions ate grass like everybody else. A strange analogy. Employers and potential employees are not really in a predator/prey relationship. (Employers and *actual* employees sometimes are, but that's a sign of a really dysfunctional business culture.) The problem is that recruiter's best interests do not align neatly with either potential employees *or* employers. They're like real estate agents. The incentives for a recruiter is to find a barely acceptable hire as quickly as possible for the least amount of effort possible. There's no point in doing extra work to find the best new hire, if the employer is willing to take a so- so hire. Since the employer is only seeing potentials that the recruiter passes on, the employer has no way of telling what the pool of would-be employees is really like. I'm not saying that all recruiters are unscrupulous or are intentionally deceiving the other parties, but the incentives are such that: - recruiters will take a bit less care to choose the right employee for the job; - they'll take a bit less care worrying about attracting the right people, because their relationship with the parties is (on average) quite short; - they're more likely The worst part of this is the vicious circle aspect. The less care recruiters put into targeting their positions, the more they get inundated with poor quality applicants. This trains applicants to carpet bomb recruiters and employers (since they're trained to expect that all job ads are misleading, and also because the unemployment office requires them to apply to X positions a week, whether X suitable positions exist or not), which gives employers an incentive to use recruiters (rather than deal with the carpet bombing of applicants). -- Steve -- https://mail.python.org/mailman/listinfo/python-list
testfixtures 4.10.0 Released!
Hi All, I'm pleased to announce the release of testfixtures 4.10.0 featuring the following: - Fixed examples in documentation broken in 4.5.1. - Add RangeComparison for comparing against values that fall in a range. - Add MockPopen.set_default(). Thanks to Asaf Peleg for the RangeComparison implementation. The package is on PyPI and a full list of all the links to docs, issue trackers and the like can be found here: https://github.com/Simplistix/testfixtures Any questions, please do ask on the Testing in Python list or on the Simplistix open source mailing list... cheers, Chris -- Simplistix - Content Management, Batch Processing & Python Consulting - http://www.simplistix.co.uk -- https://mail.python.org/mailman/listinfo/python-list
Re: Wanted Python programmer to join team
On Tuesday 17 May 2016 17:39, Steven D'Aprano wrote: > - they're more likely Er, apparently they're not more likely to do anything specific, just more likely. -- Steve -- https://mail.python.org/mailman/listinfo/python-list
Quote of the day
Overhead in the office today: "I don't have time to learn an existing library - much faster to make my own mistakes!" -- Steve -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
2016-05-17 9:50 GMT+02:00 Steven D'Aprano < steve+comp.lang.pyt...@pearwood.info>: > Overhead in the office today: > > > "I don't have time to learn an existing library - much faster to make my > own > mistakes!" > > > > -- > Steve > > -- > https://mail.python.org/mailman/listinfo/python-list > *THUMBS UP* At least they are aware of that "own mistakes" part... Not like my employer... -- https://mail.python.org/mailman/listinfo/python-list
Re: Wanted Python programmer to join team
Steven D'Aprano : > On Tuesday 17 May 2016 16:18, Marko Rauhamaa wrote: > >> Steven D'Aprano : >>> Personally, I think that advertising a job position without saying who >>> you are, what you do, and offering at least an indicative salary >>> range, are *astonishingly* rude >> I don't believe they care. >> >>> (to say nothing of counter-productive). >> Maybe, maybe not. >> >> I bet the zebras on the savannah consider the lions astonishingly rude >> and their strategy counter-productive. The savannah would be a nicer >> place if the lions ate grass like everybody else. > > A strange analogy. Employers and potential employees are not really in a > predator/prey relationship. The unsavory recruiters are the predators. > The problem is that recruiter's best interests do not align neatly > with either potential employees *or* employers. They're like real > estate agents. The incentives for a recruiter is to find a barely > acceptable hire as quickly as possible for the least amount of effort > possible. There's no point in doing extra work to find the best new > hire, if the employer is willing to take a so- so hire. Since the > employer is only seeing potentials that the recruiter passes on, the > employer has no way of telling what the pool of would-be employees is > really like. Correct. Also, they may genuinely not care about ethics of any sort. > I'm not saying that all recruiters are unscrupulous or are intentionally > deceiving the other parties, Nor am I. > but the incentives are such that: Different recruiters follow different strategies, methods and practices. Psychopathy stays in our gene pool over generations because it sometimes *is* a winning strategy. Marko -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
Radek Holý : > 2016-05-17 9:50 GMT+02:00 Steven D'Aprano < > steve+comp.lang.pyt...@pearwood.info>: > >> Overhead in the office today: >> >> "I don't have time to learn an existing library - much faster to make >> my own mistakes!" > > *THUMBS UP* At least they are aware of that "own mistakes" part... Not > like my employer... Also: With a third party solution I don't need to fix the bugs. But with an in-house solution I at least *can* fix the bugs. The feeling of powerlessness can be crushing when you depend on a third-party component that is broken with no fix in sight. Marko -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
But isn't that counter wise to batteries included? :) On Tue, May 17, 2016 at 11:30 AM, Marko Rauhamaa wrote: > Radek Holý : > > > 2016-05-17 9:50 GMT+02:00 Steven D'Aprano < > > steve+comp.lang.pyt...@pearwood.info>: > > > >> Overhead in the office today: > >> > >> "I don't have time to learn an existing library - much faster to make > >> my own mistakes!" > > > > *THUMBS UP* At least they are aware of that "own mistakes" part... Not > > like my employer... > > Also: > >With a third party solution I don't need to fix the bugs. > >But with an in-house solution I at least *can* fix the bugs. > > The feeling of powerlessness can be crushing when you depend on a > third-party component that is broken with no fix in sight. > > > Marko > -- > https://mail.python.org/mailman/listinfo/python-list > -- Sivan Greenberg Co founder & CTO Vitakka Consulting -- https://mail.python.org/mailman/listinfo/python-list
Re: OT: limit number of connections from browser to my server?
On 05/16/2016 09:31 PM, Grant Edwards wrote: On 2016-05-16, jmp wrote: Have you considered upgrading the device with a recent CPU ? Or is it completely out of the picture ? Not an option. We have to continue to support devices that are in the field. The newer models that are coming out now run at 133MHz instead of 44MHz, and page load times for https still aren't much better. That being said, your first idea seems also a good lead, have your server refuse more than one connection. Just got to figure out how to try it out without wasting a lot of time reverse-engineering the web server. Hence a Python prototype. :) I don't have time to read the whole thread but if I got it right, the main CPU consuming part is the crypto. Why not drop the https part an support only http ? Is is a device that needs to be accessed in untrusted networks ? Sorry for asking the obvious :o) jm -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
Marko Rauhamaa writes: > Radek Holý : > >> 2016-05-17 9:50 GMT+02:00 Steven D'Aprano < >> steve+comp.lang.pyt...@pearwood.info>: >> >>> Overhead in the office today: >>> >>> "I don't have time to learn an existing library - much faster to make >>> my own mistakes!" >> >> *THUMBS UP* At least they are aware of that "own mistakes" part... Not >> like my employer... > > Also: > >With a third party solution I don't need to fix the bugs. > >But with an in-house solution I at least *can* fix the bugs. > > The feeling of powerlessness can be crushing when you depend on a > third-party component that is broken with no fix in sight. > > Presumably it depends on whether you have the source for the third party component... -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
On May 17, 2016, at 4:30 AM, Marko Rauhamaa wrote: > Radek Holý : > >> 2016-05-17 9:50 GMT+02:00 Steven D'Aprano < >> steve+comp.lang.pyt...@pearwood.info>: >> >>> Overhead in the office today: >>> >>> "I don't have time to learn an existing library - much faster to make >>> my own mistakes!" >> >> *THUMBS UP* At least they are aware of that "own mistakes" part... Not >> like my employer... > > Also: > > With a third party solution I don't need to fix the bugs. > > But with an in-house solution I at least *can* fix the bugs. > > The feeling of powerlessness can be crushing when you depend on a > third-party component that is broken with no fix in sight. +1000 on this one. Just downloaded and used a library that came with unit tests, which all passed. When I started using it, I kept getting odd errors. Digging into it, I discovered they had commented out the bodies of some of the unit tests... glad it was open source, at least I *could* dig into the code and figure out what was going on :/ Thanks, Cem Karan -- https://mail.python.org/mailman/listinfo/python-list
Re: Wanted Python programmer to join team
On Tue, May 17, 2016 at 1:29 AM, Chris Angelico wrote: > On Tue, May 17, 2016 at 3:20 PM, Steven D'Aprano > wrote: >>> And a lot of job postings do come from that sort of really small >>> business, trying to expand a bit. Plus, some of them want some >>> anonymity (why, I don't know, but there are plenty of jobs posted >>> without too much in the way of company details) >> >> That probably means the job advert is coming from a recruiter. They don't >> want >> people to contact the company directly, and they want to hide the fact that >> they are a recruiter. > > Ehh, that's possible. Given that I seldom get any responses back from > them, I can't tell whether they (a) are using a recruiter who hates my > guts, (b) are using a recruiter who is utterly incompetent, (c) are > doing the recruitment themselves, and are so utterly flat-out busy > that they can't fire off a simple email to their applicants, or (d) > are doing it themselves, and don't consider failed applicants worthy > of an email. Could be any of the above, for all I can tell. Or (e), > the job was posted from just outside the Bermuda Triangle. Recruiters have changed so much in my career. Back in the early 80's you would have to have an interview with the recruiter before they would even consider submitting you for a job. They would know your skill set and they would not present a job you were not qualified for. They returned phone calls and emails. Nowadays I get emails and calls for jobs with requirements that are nowhere on my resume (e.g. .NET or sales) and when I am qualified and interested and I reply, in most cases I get no response at all. -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
On Tue, May 17, 2016 at 7:54 PM, Paul Rudin wrote: >> Also: >> >>With a third party solution I don't need to fix the bugs. >> >>But with an in-house solution I at least *can* fix the bugs. >> >> The feeling of powerlessness can be crushing when you depend on a >> third-party component that is broken with no fix in sight. >> >> > > Presumably it depends on whether you have the source for the third party > component... Yes and no. A student of mine asked me how hard it would be to use in Python a service that provided Java, .NET, and a couple of other SDKs, but not Python. Source is available for them, sure, but they're so massive and complicated that it's utterly impractical. (I had been hoping the SDKs were basically just offering a friendly API to an underlying HTTP-based service, but no.) So the options were (1) use Jython so the Java SDK became usable, or (2) fire off a subprocess that does the work and pipes it back to your app, or (3) spend about fifty years porting a gigantic lot of code to a new language. And if there'd been bugs in any of the code, well, options 1 and 2 mean my student (who knows Python but none of the SDK languages on offer) would be completely unable to fix it, source or no source - and with option 3, it'd make the port virtually impossible. Having the source available is great. It tells you which projects you DON'T want to touch. ChrisA fully aware that some of his projects will be in other people's "DON'T want to touch" boxes -- https://mail.python.org/mailman/listinfo/python-list
Re: Autotool - compile module for both Python 2 _and_ 3
Hi Ervin, On 16.05.2016 11:05, Ervin Hegedüs wrote: Hi All, there is a library, which written in C. I'ld like to use it from Python - from Python 2 _and_ 3. I can make the autotools* files for Python 2 and Python 3, but only exclusively. I can't make it for both in same time. [...] Is there any "best practice" to solve this problem? Practical advice or links are welcome! here's my practical advice: have a look at SCons (http://www.scons.org). It handles variant builds (multiple packages/target) from the same source very well, even in parallel. Best regards, Dirk -- https://mail.python.org/mailman/listinfo/python-list
Re: Autotool - compile module for both Python 2 _and_ 3
Hi Dirk, On Tue, May 17, 2016 at 09:13:02AM +0200, Dirk Bächle wrote: > Hi Ervin, > > On 16.05.2016 11:05, Ervin Hegedüs wrote: > >Hi All, > > > > > >there is a library, which written in C. I'ld like to use it from > >Python - from Python 2 _and_ 3. > > > >I can make the autotools* files for Python 2 and Python 3, but > >only exclusively. I can't make it for both in same time. > > > >[...] > > > >Is there any "best practice" to solve this problem? > > > >Practical advice or links are welcome! > > > > here's my practical advice: have a look at SCons > (http://www.scons.org). It handles variant builds (multiple > packages/target) from the same source very well, even in parallel. many thanks - I'll check it out soon. Thanks, a. -- https://mail.python.org/mailman/listinfo/python-list
Re: OT: limit number of connections from browser to my server?
Long shot here: Create a JS framework for loading resources in a better way: 1. Load HTTP and your JS core. 2. Load the rest of the resources via JS (maybe using promises for chaining the requests one after the other) -- https://mail.python.org/mailman/listinfo/python-list
EuroPython 2016 Keynote: Nicholas Tollervey
We are pleased to introduce our first keynote speaker for EuroPython 2016: *** Nicholas Tollervey *** About Nicholas Tollervey Nicholas is a classically trained musician, philosophy graduate, teacher, author (for O'Reilly) and freelance programmer. He was a founding member of the London Python Code Dojo, created and organized PyCon UK’s education track and is a fellow of the Python Software Foundation. He also proposed, coordinated and contributed to the PSF’s partnership with the BBC in the creation of the micro:bit. He’s just like this biography: concise, honest and full of useful information. The Keynote: A Million Children (and MicroPython) - The BBC micro:bit is a small programmable device for children. A million of them have been handed out to the UK’s 11 and 12 years olds. The Python Software Foundation was a partner in this moon-shot scheme and, thanks to the efforts of Damien George, MicroPython (a full reimplementation of Python 3) runs on the device: https://en.wikipedia.org/wiki/Micro_Bit http://micropython.org/ In his keynote, Nicholas will tell the story of the project, describe Python’s role in it and explain how the wider Python community can become involved. It may involve demonstrations, live coding and audience participation. With gravitational regards, -- EuroPython 2016 Team http://ep2016.europython.eu/ http://www.europython-society.org/ -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
Paul Rudin : > Marko Rauhamaa writes: >> The feeling of powerlessness can be crushing when you depend on a >> third-party component that is broken with no fix in sight. > > Presumably it depends on whether you have the source for the third > party component... Just having such an experience. The linux kernel has a critical bug in a major distribution (who shall be left unnamed here) that has been fixed in a later kernel version. Thanks to linux being free software, I managed to pin down the root cause after more than a month of debugging. I sent a bug report to the linux vendor and attached a tiny patch. The vendor has graciously agreed to consider releasing an update in the summer (we are in the process of verifying the fix). The problem was first detected in December. A semi-reliable reproduction was discovered in early February. The root cause and proposed fix was identified mid-March. A vendor fix will likely come out by the end of June. That's a long time to be without a product to sell. Marko -- https://mail.python.org/mailman/listinfo/python-list
Re: OT: limit number of connections from browser to my server?
On 2016-05-17, Gregory Ewing wrote: > Is there some way you can get more stuff into a single > html page? For example, use inline css and image data > instead of delivering them as separate files. Yes. That's one option that's still on the table, and that's probably what the smart money is betting on: use server-side includes for css and javascript files. For image data, I'd need to add a server-side-include-and-base64-encode capability, but that shouldn't take more than a couple hours. Then the HTML editing starts... I've seen examples of including base64 image data in HTML tags, but don't know how to do it in CSS for things like backgrounds. That said, I've got two more things to try to see if I can convince browsers to behave: 1) Stall the connection after the TCP connection opens but before the SSL handshake. 2) Try listen(0) on my BSD-derived network stack to limit the number of TCP connections. IIRC, BSD behaves differently than Linux when the listen queue overflows. I doubt either one will help -- and even if they do, it'll probably all fall apart again when new browser versions come out. I'm also looking into SSL session resumption. My SSL stack appears to support it, but I can't tell if it's being used or not by browsers. It may be a moot point. All of the descriptions I've read about SSL session resumption say it's for low-overhead reopening of a previously closed connection. That's not what I need: I need to speed up opening of a new connection while the first connection is still open. -- Grant Edwards grant.b.edwardsYow! Mary Tyler Moore's at SEVENTH HUSBAND is wearing gmail.commy DACRON TANK TOP in a cheap hotel in HONOLULU! -- https://mail.python.org/mailman/listinfo/python-list
Re: OT: limit number of connections from browser to my server?
On 2016-05-17, Marko Rauhamaa wrote: > Gregory Ewing : > >> Is there some way you can get more stuff into a single >> html page? For example, use inline css and image data >> instead of delivering them as separate files. > > Better yet, is there some way you could send less stuff? No. Somebody else decides what the pages look like and behaves. I don't get to toss out logos, background images, customized checkbox glyphs, etc. I get to make it work. :) -- Grant Edwards grant.b.edwardsYow! I have a very good at DENTAL PLAN. Thank you. gmail.com -- https://mail.python.org/mailman/listinfo/python-list
Re: OT: limit number of connections from browser to my server?
On 2016-05-17, Rodrigo Bistolfi wrote: > Long shot here: Create a JS framework for loading resources in a better way: > > 1. Load HTTP and your JS core. > > 2. Load the rest of the resources via JS (maybe using promises for >chaining the requests one after the other) I thought about that. It's probably more work than going the "single-fetch" route via server-side includes, and probably more fragile. -- Grant Edwards grant.b.edwardsYow! An air of FRENCH FRIES at permeates my nostrils!! gmail.com -- https://mail.python.org/mailman/listinfo/python-list
Re: OT: limit number of connections from browser to my server?
On 2016-05-17, jmp wrote: > I don't have time to read the whole thread but if I got it right, > the main CPU consuming part is the crypto. Yep. > Why not drop the https part an support only http ? Product spec explicitly states HTTPS only. I'm told that is not open for discussion. The customer is a large, somewhat bureaucratic German corporation, and they generally mean it when they say something is non-negotiable. > Is is a device that needs to be accessed in untrusted networks? > Sorry for asking the obvious :o) -- Grant Edwards grant.b.edwardsYow! I'm EMOTIONAL at now because I have gmail.comMERCHANDISING CLOUT!! -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
Marko Rauhamaa writes: > Paul Rudin : > >> Marko Rauhamaa writes: >>> The feeling of powerlessness can be crushing when you depend on a >>> third-party component that is broken with no fix in sight. >> >> Presumably it depends on whether you have the source for the third >> party component... > > Just having such an experience. The linux kernel has a critical bug in a > major distribution (who shall be left unnamed here) that has been fixed > in a later kernel version. > > Thanks to linux being free software, I managed to pin down the root > cause after more than a month of debugging. I sent a bug report to the > linux vendor and attached a tiny patch. The vendor has graciously agreed > to consider releasing an update in the summer (we are in the process of > verifying the fix). > > The problem was first detected in December. A semi-reliable reproduction > was discovered in early February. The root cause and proposed fix was > identified mid-March. A vendor fix will likely come out by the end of > June. > > That's a long time to be without a product to sell. > But you do have the option of building a kernel incorporating your fix and using that. -- https://mail.python.org/mailman/listinfo/python-list
Re: Design: Idiom for classes and methods that are customizable by the user?
Hi Michael, and thanks a lot for chiming in on this topic. On 13.05.2016 22:33, Michael Selik wrote: [...] I share Greg's trepidation when I hear a phrase like that, but the general idea of a registry of classes or functions and then picking the right one based on string input is fine. See my answer to Gregory, addressing his concerns. ;) How would the API work for custom Taskmasters? It's not so great to require that the user must explicitly ``add`` their derived class after defining it. Perhaps that add function could be a decorator? Our current API doesn't use decorators at all, since it's also aimed at people with no (or only some) knowledge of Python. Until a year or so ago, we have been backwards compatible down to Python 1.5.2...and then to Python 2.4 for some time. Only during the last two months we made a clear cut and are now seeing Python 2.7 as our floor to further develop from. So to some degree, decorators were technically not possible to support...and would have looked strange to the average user perhaps? We're now free to add them into the mix for our new Python 2.7.x/3.y codebase, but this would probably require an API discussion first. However, I'll keep the "decorator" idea in mind when I get back to putting "all the dots on the i" for the interface. Best regards, Dirk -- https://mail.python.org/mailman/listinfo/python-list
Re: How to create development Python environment on Linux.
On Mon, May 16, 2016 at 4:28 PM, wrote: > Thanks Zach, that's a big help. The only reason I want to get a Python 2.7 > environment working first is because I'll be working on third party code and > that's the platform it uses. For any new projects I would use Python 3. Fair enough :) > After considering your guidance I think what I will do is install virtualenv > using apt-get and then use that to create a dev environment. Is it ok to run > get-pip.py in a virtual environment? Sounds like a plan. get-pip.py should work fine in a venv, but first check to see if pip is already there, as Michael mentioned. > I won't worry about using the latest version of 2.7 for now, since it's only > one or two third party open source projects I'll use 2.7 for and they don't > need a more recent version. > > There are a couple of other things I'm not quite clear on, such as where it > would be best to create my new virtual environment (I'm tempted to put it in > /usr/local if that means it can be used by all user accounts on my machine), > and how I can can control which Python environment is used by the various > system and user programs that depend on them, but I expect I can find that > information on the web, though I'll make another post here if I do get stuck. The thing about virtual environments is that you can create as many as you want wherever you want, and they're all independent of each other. All you need to do is have a way to consistently recreate the same environment, which is usually easiest to do with a 'requirements.txt' file with all of your dependencies' versions pinned. Then recreating the venv is as simple as `virtualenv --python=some_python /path/to/venv && /path/to/venv/bin/pip install -r /path/to/requirements.txt`. Then you can do your development wherever you want and install into somewhere in `/usr/local` when you deploy. For example, one project I work on creates a venv in `/usr/local/lib/project_name/`, then creates links to the entry point scripts (that actually live in `/usr/local/lib/project_name/bin/`) in `/usr/local/bin/`. As for controlling which environment is used: all system scripts/apps should specify some version-specific flavor of /usr/bin/python in their shebang--if they don't it's an OS bug and should be filed as such. For your stuff, the path of least headache is probably to create a venv for each one, and specify the path to the venv's python in shebang lines. Hope this helps, -- Zach -- https://mail.python.org/mailman/listinfo/python-list
Re: OT: limit number of connections from browser to my server?
Steven D'Aprano wrote: > On Tue, 17 May 2016 02:52 am, Chris Angelico wrote: > >> On Tue, May 17, 2016 at 2:34 AM, Rob Gaddi >> wrote: The solution might actually be to move all your static files elsewhere. Slap 'em up onto github.io or something, and then the browser is free to make all the parallel connections it likes; your embedded device can just serve the stuff that actually varies (presumably the main HTML file). I know that isn't what you asked for, but it's something to consider :) ChrisA >>> >>> Oooof. Not to be rude, Chris, but your "software guy" is showing. >>> Grant's got the right of it; if you're shipping a box with an RJ-45 and >>> a webpage, and you want the customer to be able to always make it >>> work, then it needs to be a self-contained entity. The belief that your >>> external dependancies will always be there is why leftpad was able to >>> break everything, and why Google just bricked a bunch of people's >>> expensive Revolv Hubs. > > Schadenfreude is a beautiful emotion :-) > > "Yes, let's put a critical requirement of our business in the hands of a > third party with absolutely *no* obligations to us, and no government > oversight. What could *possibly* go wrong???" > > >> I agree, but I also make no apology for suggesting the option of >> getting someone else to do some of the work. In this case, it can be >> rejected for the exact reason you cite (dependencies are a cost, and >> in this case way too high a cost), and that's fine and correct. >> Ultimately, if the job gets done, everything else is implementation >> detail, with consequences - and I know a lot of people who'll >> willingly sacrifice "reliability in the face of an internet connection >> outage" in favour of "less than fifteen second response time". > > How can you not serve a web page over your LAN in 15s? > > I mean, you could *almost* do it by hand, copying the files onto a USB stick > and walking them across the room in 15 seconds. Maybe 30. > Simple, because embedded web servers running on toy microprocessors are HARD. When you're trying to work with whatever browser the customer may have, you have no control over the number of simultaneous connections it will try to make to your box. If you don't allow those connections it can cause huge stalls by forcing the TCP layer to time out. If you do, in Grant's case, it forces you to perform tons of expensive public-key crypto on a 40 MHz processor (which, hmmm, external memory bus, ~40 MHz... Coldfire?). A lot of things you can take for granted on a compute monster (like a Chromebook or Atom based laptop) get much more complicated when you're resource constrained. -- Rob Gaddi, Highland Technology -- www.highlandtechnology.com Email address domain is currently out of order. See above to fix. -- https://mail.python.org/mailman/listinfo/python-list
Re: Design: Idiom for classes and methods that are customizable by the user?
On Tue, May 17, 2016 at 11:18 AM Dirk Bächle wrote: > > > It's not so great to require > > that the user must explicitly ``add`` their derived class after defining > > it. Perhaps that add function could be a decorator? > > Our current API doesn't use decorators at all, since it's also aimed at > people with no (or only some) knowledge of Python. I think you'll find that users of decorators do not need any special knowledge. The Flask framework (http://flask.pocoo.org/) is proof of that. > decorators...would have looked strange to the average user perhaps? > Many beginners are willing to mimic your example code without questioning what that "@" thingy is doing above the function or class. Compare that with trying to explain to a C++ user that in Python classes are actually objects and you can pass them to the ``add`` function to register them... that's going to take a while. We're now free to add them into the mix for our new Python 2.7.x/3.y > codebase, but this would probably require an API discussion > first. > An alternative would be creating a TaskmasterMeta metaclass that registers the Taskmaster subclass when it's defined. I'm a little reluctant to recommend that one. I find decorators to be more elegant. -- https://mail.python.org/mailman/listinfo/python-list
Re: OT: limit number of connections from browser to my server?
On 2016-05-17, Rob Gaddi wrote: > >> How can you not serve a web page over your LAN in 15s? >> >> I mean, you could *almost* do it by hand, copying the files onto a >> USB stick and walking them across the room in 15 seconds. Maybe 30. > > Simple, because embedded web servers running on toy microprocessors are > HARD. 40MHz with multiple MB of RAM is pretty high-end in my book. I've worked on projects where the clock speed was in KHz, the supply current was measrued in micro-Amps, and there were a few KB of RAM. Of course you don't try to do TCP and SSL on something like that. > When you're trying to work with whatever browser the customer may > have, you have no control over the number of simultaneous connections > it will try to make to your box. If you don't allow those connections > it can cause huge stalls by forcing the TCP layer to time out. If you > do, in Grant's case, it forces you to perform tons of expensive > public-key crypto on a 40 MHz processor (which, hmmm, external memory > bus, ~40 MHz... Coldfire?). The 40MHz one is a Samsung ARM7TDMI. There's a newer model with a 133MHz Cortex-M3. For most things it's 2-3 times faster than the ARM7, but the ARM7 has an I/D cache and the M3 doesn't. So there are few highly localized tasks where there's not a lot of difference. > A lot of things you can take for granted on a compute monster (like > a Chromebook or Atom based laptop) get much more complicated when > you're resource constrained. And that's what keeps me paid. :) -- Grant Edwards grant.b.edwardsYow! I know th'MAMBO!! at I have a TWO-TONE CHEMISTRY gmail.comSET!! -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
On 05/17/2016 08:27 AM, Paul Rudin wrote: > Marko Rauhamaa writes: >> That's a long time to be without a product to sell. > > But you do have the option of building a kernel incorporating your fix > and using that. Sure as an individual end user that may be the best option. But not necessarily for a business. The cost of doing that could be prohibitive. Sometimes we forget just how costly open source software can be (really *all* software). They can either deal with lost revenue waiting, or they can budget a tremendous amount of money, time, and effort to support their own kernel which would entail doing updates, QA testing, etc. Letting the upstream vendor do all that (their core business after all) is often the least costly option. Though it sounds like they've already spent a lot of money doing QA to identify this bug. When I did IT professionally, our policy with regards to Linux was to stick with existing packages from a known set of (mostly) official channels and to discourage any installing of libraries and frameworks from source. Allowing packages to be installed from source was just a maintenance nightmare. RPM (or deb or whatever) brings at least a tiny bit of stability and consistency. -- https://mail.python.org/mailman/listinfo/python-list
Re: Quote of the day
Michael Torrie : > On 05/17/2016 08:27 AM, Paul Rudin wrote: >> Marko Rauhamaa writes: >>> That's a long time to be without a product to sell. >> >> But you do have the option of building a kernel incorporating your fix >> and using that. > > Sure as an individual end user that may be the best option. But not > necessarily for a business. Correct, the answer would be no. > The cost of doing that could be prohibitive. Really, the customer would simply refuse to do it. They are not in the business of building kernels. Also, they would immediately fall out of any kind of distro support if they improvised with their own kernel. > When I did IT professionally, our policy with regards to Linux was to > stick with existing packages from a known set of (mostly) official > channels and to discourage any installing of libraries and frameworks > from source. Allowing packages to be installed from source was just a > maintenance nightmare. RPM (or deb or whatever) brings at least a tiny > bit of stability and consistency. Exactly. Marko -- https://mail.python.org/mailman/listinfo/python-list
Python 3.6.0a1 is now available
On behalf of the Python development community and the Python 3.6 release team, I'm happy to announce the availability of Python 3.6.0a1. 3.6.0a1 is the first of four planned alpha releases of Python 3.6, the next major release of Python. During the alpha phase, Python 3.6 remains under heavy development: additional features will be added and existing features may be modified or deleted. Please keep in mind that this is a preview release and its use is not recommended for production environments. You can find Python 3.6.0a1 here: https://www.python.org/downloads/release/python-360a1/ The next release of Python 3.6 will be 3.6.0a2, currently scheduled for 2016-06-13. Enjoy! --Ned -- Ned Deily n...@python.org -- [] -- https://mail.python.org/mailman/listinfo/python-list
Backwards-incompatible changes
Every change breaks somebody's workflow. http://xkcd.com/1172/ -- Steve -- https://mail.python.org/mailman/listinfo/python-list
