Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #76962 > unrolled thread

Re: GO vs Python

Back to article view | Back to comp.lang.python

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: GO vs Python Chris Angelico <rosuav@gmail.com> - 2014-08-25 13:41 +1000
    Re: GO vs Python Rustom Mody <rustompmody@gmail.com> - 2014-08-25 01:01 -0700
      Re: GO vs Python Marko Rauhamaa <marko@pacujo.net> - 2014-08-25 11:52 +0300
      Re: GO vs Python Chris Angelico <rosuav@gmail.com> - 2014-08-25 19:48 +1000
      Re: GO vs Python Roy Smith <roy@panix.com> - 2014-08-25 08:06 -0400
        Re: GO vs Python Rustom Mody <rustompmody@gmail.com> - 2014-08-25 05:24 -0700
          Re: GO vs Python Michael Torrie <torriem@gmail.com> - 2014-08-25 08:13 -0600
            Re: GO vs Python Marko Rauhamaa <marko@pacujo.net> - 2014-08-25 17:26 +0300
              Re: GO vs Python Rustom Mody <rustompmody@gmail.com> - 2014-08-25 07:38 -0700
                Re: GO vs Python Chris “Kwpolska” Warrick <kwpolska@gmail.com> - 2014-08-25 16:44 +0200
                  Re: GO vs Python Marko Rauhamaa <marko@pacujo.net> - 2014-08-25 18:04 +0300
                    Re: GO vs Python Mark Lawrence <breamoreboy@yahoo.co.uk> - 2014-08-25 17:40 +0100
          Re: GO vs Python Olaf Hering <olaf@aepfle.de> - 2014-08-25 16:32 +0200
          Re: GO vs Python Roy Smith <roy@panix.com> - 2014-08-25 21:21 -0400

#76962 — Re: GO vs Python

On Mon, Aug 25, 2014 at 1:27 PM, Sam Fourman Jr. <sfourman@gmail.com> wrote:
> I remember doing some browsing around, and the pooco people that make jinja2
> were not fans of python3(I forget the blog post), I got scared because a
> very large portion of my income was based on a single client... So since we
> were having scalability issues anyway, I moved them to GO, and it was a Win
> - Win, the GO standard lib does so much, and the scalability gains we
> received over python were so large, that we were able to reduce out AWS bill
> so much that I could hire another coder.

There are some communities that, for some reason or other, dislike
Python 3. That doesn't mean you have to. The Py2 model is a bit easier
for boundary code (it lets you stuff your fingers in your ears and go
"LALALALALA there are no character encodings"), but the Py3 model is
way easier for application code. Text is text, no matter what
characters it has in it.

> I really like python, and we use it a ton, but a python like compiled
> language did wonders for us when we needed it most.

Sure. And your reduction of AWS bills sounds great. Just make sure you
don't consume the entire extra coder's time doing things that Python
would have done for you. Go's character model is inferior to Python
3's (or at least, it was last time I checked - stuff might have
changed since then), so you may find yourself doing a lot of
unnecessary work to make sure your code works correctly. Do be sure to
test everything thoroughly, with characters from all over Unicode.

Personally, when I want "Python but faster", I go to Pike. Same
character/string model (even the same style of internal
representation), same broad object model, but a stronger focus on
networking and on staying running 100% of the time.

ChrisA

[toc] | [next] | [standalone]

#76969

On Monday, August 25, 2014 9:11:39 AM UTC+5:30, Chris Angelico wrote:

> Sure. And your reduction of AWS bills sounds great. Just make sure you
> don't consume the entire extra coder's time doing things that Python
> would have done for you. Go's character model is inferior to Python
> 3's (or at least, it was last time I checked - stuff might have
> changed since then), so you may find yourself doing a lot of
> unnecessary work to make sure your code works correctly. Do be sure to
> test everything thoroughly, with characters from all over Unicode.

Heh! You make it sound that the character model is the most important thing
in choosing a language!
There are people using Fortran -- with not intention of finding
an alternative.

[toc] | [prev] | [next] | [standalone]

#76970

Rustom Mody <rustompmody@gmail.com>:

> Heh! You make it sound that the character model is the most important
> thing in choosing a language!

That's because the "character model" is the raison-d'être for Python3.

As far as Go goes, I think it's an interesting approach, but marred by
the Google tie-in. You must have a Google account to participate in the
Go community.


Marko

[toc] | [prev] | [next] | [standalone]

#76971

On Mon, Aug 25, 2014 at 6:01 PM, Rustom Mody <rustompmody@gmail.com> wrote:
> On Monday, August 25, 2014 9:11:39 AM UTC+5:30, Chris Angelico wrote:
>
>> Sure. And your reduction of AWS bills sounds great. Just make sure you
>> don't consume the entire extra coder's time doing things that Python
>> would have done for you. Go's character model is inferior to Python
>> 3's (or at least, it was last time I checked - stuff might have
>> changed since then), so you may find yourself doing a lot of
>> unnecessary work to make sure your code works correctly. Do be sure to
>> test everything thoroughly, with characters from all over Unicode.
>
> Heh! You make it sound that the character model is the most important thing
> in choosing a language!
> There are people using Fortran -- with not intention of finding
> an alternative.

Heh. Well, in some ways, it is. When you advise someone on how to
choose a car, do you say "Make sure you get one with wheels"? No. You
assume it. There are far more important aspects to a programming
language, but most of them are fairly obvious (and often universal).
Also, if someone's trying to decide between C (manual memory
management, compiled, detaily) and Python (garbage collected, high
level), he'll know the issues. I'm referring here to something that
might be a surprise way down the track.

Let's take a concrete example. You want to build a database-backed
dynamic web site. You could use PHP and MySQL, and a few years ago
that would be your only option if you want cheap web hosting. Or you
could use Python and PostgreSQL, tying in with Passenger to manage
stuff under Apache. Or you could write an entire HTTP server in Python
(using http.server or BaseHTTPServer, perhaps), putting it on a
dedicated IP or port. Or you could probably craft something horrendous
out of bash scripts, massive pipelines, and some command-line tool
that binds to a port (I know you can use /dev/tcp/hostname/port to
connect, not sure how you'd listen). If someone asks you for advice,
what aspects of each would you highlight? Obvious stuff doesn't need
to be mentioned (like "PHP uses braces, if you hate braces then use
Python", and "doing it all with bash pipelines is Just Plain Dumb"),
and what you need to say is stuff like "Doing it with http.server will
limit you to one CPU core" and "Doing it with PHP means you don't get
Unicode support".

So yes, I think it was the most important thing to mention, because
it's something that isn't obvious to a lot of people. You test your
code with stuff you can type, and it works fine. Other people use
non-ASCII characters, and nearly everything works, but there are just
these occasional glitches (because your code's working with UTF-8
bytes, and every once in a while it counts things differently or
something). Working with bytes instead of characters is one of the
most subtle problems to infect a program, because 99%+ of it works
fine. And properly fixing it is a huge job. That's why it's worth
saying, even if character model isn't actually the one most important
feature of a programming language.

ChrisA

[toc] | [prev] | [next] | [standalone]

#76976

In article <d8811b7e-fc37-447c-848d-f3bd314aff5e@googlegroups.com>,
 Rustom Mody <rustompmody@gmail.com> wrote:

> Heh! You make it sound that the character model is the most important thing
> in choosing a language!
> There are people using Fortran -- with not intention of finding
> an alternative.

Different people have different needs.  If I was writing code to do 
number crunching, I wouldn't care much if I could only print 7-bit 
ascii.  If I was writing code to serve music on the web and had to 
display artist names like Mötley Crüe and Beyoncé, I would be more 
concerned about the character model.

[toc] | [prev] | [next] | [standalone]

#76977

On Monday, August 25, 2014 5:36:25 PM UTC+5:30, Roy Smith wrote:
>  Rustom Mody  wrote:

> > Heh! You make it sound that the character model is the most important thing
> > in choosing a language!
> > There are people using Fortran -- with not intention of finding
> > an alternative.

> Different people have different needs.  If I was writing code to do 
> number crunching, 

Precisely:
- Compiled language + fast compilation
- Static binaries -- better (in the sense of more lo-level) than C's default
- Static strongly typed + type-inference
- Interfaces instead of inheritance
- And of course "I want a google-job"

and probably many others can be the deciders for many people

> ... I wouldn't care much if I could only print 7-bit ascii.
> If I was writing code to serve music on the web and had to 
> display artist names like M�tley Cr�e and Beyonc�, I would be more 
> concerned about the character model.

Funny!!
Your mail client seems to be 7-bit ASCII!!

[toc] | [prev] | [next] | [standalone]

#76981

On 08/25/2014 06:24 AM, Rustom Mody wrote:
>> ... I wouldn't care much if I could only print 7-bit ascii.
>> If I was writing code to serve music on the web and had to 
>> display artist names like M�tley Cr�e and Beyonc�, I would be more 
>> concerned about the character model.
> 
> Funny!!
> Your mail client seems to be 7-bit ASCII!!

No it came through fine here, originally.  It must be Google Groups that
messed up the characters in displaying the message and in your reply.

[toc] | [prev] | [next] | [standalone]

#76982

Michael Torrie <torriem@gmail.com>:

> On 08/25/2014 06:24 AM, Rustom Mody wrote:
>>> display artist names like M�tley Cr�e and Beyonc�, I would be more 
>> 
>> Your mail client seems to be 7-bit ASCII!!
>
> No it came through fine here, originally. It must be Google Groups
> that messed up the characters in displaying the message and in your
> reply.

I saw it fine (<URL: news:comp.lang.python>), but there were no MIME
headers. MIME might not be part of the NNTP standard, but is a good idea
even there. GNUS must have guessed the encoding based on my locale.


Marko

[toc] | [prev] | [next] | [standalone]

#76983

On Monday, August 25, 2014 7:56:55 PM UTC+5:30, Marko Rauhamaa wrote:
> Michael Torrie:

> > On 08/25/2014 06:24 AM, Rustom Mody wrote:
> >>> display artist names like M�tley Cr�e and Beyonc�, I would be more 
> >> Your mail client seems to be 7-bit ASCII!!
> > No it came through fine here, originally. It must be Google Groups
> > that messed up the characters in displaying the message and in your
> > reply.

> I saw it fine (<URL: news:comp.lang.python>), but there were no MIME
> headers. MIME might not be part of the NNTP standard, but is a good idea
> even there. GNUS must have guessed the encoding based on my locale.

> Marko

Mysterious...

This page 
https://mail.python.org/pipermail/python-list/2014-August/677408.html
shows charset as us-ascii
And it shows Mötley Crüe and Beyoncé
So it means ö ü é are us-ascii?

[Of course I understand none of all this]

रुसि [To force GG to unicode this message]

[toc] | [prev] | [next] | [standalone]

#76984

On Mon, Aug 25, 2014 at 4:38 PM, Rustom Mody <rustompmody@gmail.com> wrote:
> On Monday, August 25, 2014 7:56:55 PM UTC+5:30, Marko Rauhamaa wrote:
>> Michael Torrie:
>
>> > On 08/25/2014 06:24 AM, Rustom Mody wrote:
>> >>> display artist names like M�tley Cr�e and Beyonc�, I would be more
>> >> Your mail client seems to be 7-bit ASCII!!
>> > No it came through fine here, originally. It must be Google Groups
>> > that messed up the characters in displaying the message and in your
>> > reply.
>
>> I saw it fine (<URL: news:comp.lang.python>), but there were no MIME
>> headers. MIME might not be part of the NNTP standard, but is a good idea
>> even there. GNUS must have guessed the encoding based on my locale.
>
>> Marko
>
> Mysterious...
>
> This page
> https://mail.python.org/pipermail/python-list/2014-August/677408.html
> shows charset as us-ascii
> And it shows Mötley Crüe and Beyoncé
> So it means ö ü é are us-ascii?
>
> [Of course I understand none of all this]
>
> रुसि [To force GG to unicode this message]
> --
> https://mail.python.org/mailman/listinfo/python-list

Your browser helpfully ignores that header and uses ISO-8859-1, which
*is* the actual encoding of that message.  Gmail displays the message
properly me, but apparently Google Groups defaults to UTF-8, and that
results in question marks.

-- 
Chris “Kwpolska” Warrick <http://chriswarrick.com/>
PGP: 5EAAEA16
stop html mail | always bottom-post | only UTF-8 makes sense

[toc] | [prev] | [next] | [standalone]

#76987

Chris “Kwpolska” Warrick <kwpolska@gmail.com>:

>>> >>> display artist names like M�tley Cr�e and Beyonc�, I would be more
>>> >> Your mail client seems to be 7-bit ASCII!!
>>> > No it came through fine here, originally.
>>
>>> I saw it fine (<URL: news:comp.lang.python>), but there were no MIME
>>> headers. MIME might not be part of the NNTP standard, but is a good idea
>>> even there. GNUS must have guessed the encoding based on my locale.
>>
>> https://mail.python.org/pipermail/python-list/2014-August/677408.html
>> shows charset as us-ascii
>> And it shows Mötley Crüe and Beyoncé
>> So it means ö ü é are us-ascii?
>
> Your browser helpfully ignores that header and uses ISO-8859-1, which
> *is* the actual encoding of that message.  Gmail displays the message
> properly me, but apparently Google Groups defaults to UTF-8, and that
> results in question marks.

Thank goodness for Python3, which has freed us from these archaic
confusions.


Marko

[toc] | [prev] | [next] | [standalone]

#76991

On 25/08/2014 16:04, Marko Rauhamaa wrote:
>
> Thank goodness for Python3, which has freed us from these archaic
> confusions.
>

It had to happen.  Cue our resident unicode expert who will tell us yet 
again how fatally flawed the FSR is but once again fail to explain why. 
  Hum, which microbenchmark will get trotted out this time?

-- 
My fellow Pythonistas, ask not what our language can do for you, ask
what you can do for our language.

Mark Lawrence

[toc] | [prev] | [next] | [standalone]

#76985

On Mon, Aug 25, Michael Torrie wrote:

> No it came through fine here, originally.  It must be Google Groups that
> messed up the characters in displaying the message and in your reply.

Likely not. The header I got with msgid
<roy-592146.08040025082014@news.panix.com> lacks basic headers like
"Content-Type:" and "Content-Transfer-Encoding:". Funny that this
happend in a thread about encoding...


Olaf

[toc] | [prev] | [next] | [standalone]

#77015

In article <cb2aff1b-c091-48b5-90f5-d82c6060b148@googlegroups.com>,
 Rustom Mody <rustompmody@gmail.com> wrote:

> On Monday, August 25, 2014 5:36:25 PM UTC+5:30, Roy Smith wrote:
> >  Rustom Mody  wrote:
> 
> > > Heh! You make it sound that the character model is the most important 
> > > thing
> > > in choosing a language!
> > > There are people using Fortran -- with not intention of finding
> > > an alternative.
> 
> > Different people have different needs.  If I was writing code to do 
> > number crunching, 
> 
> Precisely:
> - Compiled language + fast compilation
> - Static binaries -- better (in the sense of more lo-level) than C's default
> - Static strongly typed + type-inference
> - Interfaces instead of inheritance
> - And of course "I want a google-job"
> 
> and probably many others can be the deciders for many people
> 
> > ... I wouldn't care much if I could only print 7-bit ascii.
> > If I was writing code to serve music on the web and had to 
> > display artist names like M?tley Cr?e and Beyonc?, I would be more 
> > concerned about the character model.
> 
> Funny!!
> Your mail client seems to be 7-bit ASCII!!

Well, not my mail client.  My NNTP client.  I'm running a hideously 
outdated client (MT-Newswatcher) because I love the U/I.  Unfortunately, 
it's pretty braindead in the unicode department.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web