librelist archives

« back to archive

Re: [flask] Web Framework shootout

Re: [flask] Web Framework shootout

From:
Date:
2011-04-13 @ 18:53
Hey Jonathan,

I think the whole point to Flask being a micro framework is so that anyone
can easily modify it to suit their needs. Out of the box it's fast and 
lean. This works for most people and you won't notice any speed issues. 
If you need Flask to run a service like Twitter, you have quite a lot of 
work ahead of you. But any large web app will require substantial 
deviation from the original framework, whether homebrew or not.
Sent from my BlackBerry device on the Rogers Wireless Network

-----Original Message-----
From: Jonathan Chen <tamasiaina@gmail.com>
Sender: flask@librelist.com
Date: Wed, 13 Apr 2011 11:32:36 
To: <flask@librelist.com>
Reply-To: flask@librelist.com
Subject: [flask] Web Framework shootout

http://blog.curiasolutions.com/the-great-web-framework-shootout/

Has anybody read this article? It seems like bottle and flask do pretty well
compared to other frameworks when dealing with templates and a database. Now
I know that these articles are not really containing the full picture, but
is there any way we can try to make flask faster and optimize it?


~Jonathan C.

Re: [flask] Web Framework shootout

From:
dag.odenhall@gmail.com
Date:
2011-04-13 @ 19:40
I would say that Flask values elegance of API and source code over raw
performance, when the gains in performance would be negligible. That's
not to say Flask is slow, though.

On 13 April 2011 20:53,  <mikecampo@gmail.com> wrote:
> Hey Jonathan,
>
> I think the whole point to Flask being a micro framework is so that anyone
> can easily modify it to suit their needs. Out of the box it's fast and lean.
> This works for most people and you won't notice any speed issues.
> If you need Flask to run a service like Twitter, you have quite a lot of
> work ahead of you. But any large web app will require substantial deviation
> from the original framework, whether homebrew or not.
>
> Sent from my BlackBerry device on the Rogers Wireless Network
>
> ________________________________
> From: Jonathan Chen <tamasiaina@gmail.com>
> Sender: flask@librelist.com
> Date: Wed, 13 Apr 2011 11:32:36 -0700
> To: <flask@librelist.com>
> ReplyTo: flask@librelist.com
> Subject: [flask] Web Framework shootout
> http://blog.curiasolutions.com/the-great-web-framework-shootout/
>
> Has anybody read this article? It seems like bottle and flask do pretty well
> compared to other frameworks when dealing with templates and a database. Now
> I know that these articles are not really containing the full picture, but
> is there any way we can try to make flask faster and optimize it?
>
>
> ~Jonathan C.
>

Re: [flask] Web Framework shootout

From:
Am
Date:
2011-04-13 @ 23:15
If you look at the y scale (requests per second) against flask and bottle.
Bottle drops over the three tests from around 3100 req/s to 1800 req/s and
Flask drops from 2100 req/s to 1400 req/s.

Over all Flask drops by 700 req/s over the three tests, whereas Bottle drops
1300 req/s over the three tests.

If you go further. By test Flask drops 2100 > 1900 (200 req/s) from the
hello world test to the templating test. Then 1900 > 1400 (500 req/s) from
the templating test to the template/db test.

wonder what his database code looks like or if it could be improved, im
guessing it probably can.

Bottle is a little more consistent doping 3100 > 2500 (600 req/s) then 2500
> 1800 (700 req/s)

overall bottle does drop pretty heavily as soon as you move from basic hello
world apps but it doesnt look that obvious in the graphics he provides.

The bigger frameworks don't do much better doping by the thousand each time.

It would be nice if his graphics actually had a consistent scale.


On 13 April 2011 19:40, dag.odenhall@gmail.com <dag.odenhall@gmail.com>wrote:

> I would say that Flask values elegance of API and source code over raw
> performance, when the gains in performance would be negligible. That's
> not to say Flask is slow, though.
>
> On 13 April 2011 20:53,  <mikecampo@gmail.com> wrote:
> > Hey Jonathan,
> >
> > I think the whole point to Flask being a micro framework is so that
> anyone
> > can easily modify it to suit their needs. Out of the box it's fast and
> lean.
> > This works for most people and you won't notice any speed issues.
> > If you need Flask to run a service like Twitter, you have quite a lot of
> > work ahead of you. But any large web app will require substantial
> deviation
> > from the original framework, whether homebrew or not.
> >
> > Sent from my BlackBerry device on the Rogers Wireless Network
> >
> > ________________________________
> > From: Jonathan Chen <tamasiaina@gmail.com>
> > Sender: flask@librelist.com
> > Date: Wed, 13 Apr 2011 11:32:36 -0700
> > To: <flask@librelist.com>
> > ReplyTo: flask@librelist.com
> > Subject: [flask] Web Framework shootout
> > http://blog.curiasolutions.com/the-great-web-framework-shootout/
> >
> > Has anybody read this article? It seems like bottle and flask do pretty
> well
> > compared to other frameworks when dealing with templates and a database.
> Now
> > I know that these articles are not really containing the full picture,
> but
> > is there any way we can try to make flask faster and optimize it?
> >
> >
> > ~Jonathan C.
> >
>

Re: [flask] Web Framework shootout

From:
Col Wilson
Date:
2011-04-14 @ 06:38
I came to flask from bottle. Bottle is great, but I wanted just a little bit
more than it had to offer at the time and flask had it. I'll use bottle
again I'm sure, but those are the decisions there for us to make.

Facts (and I don't see any particular reasons to doubt the results) are
there to advise us to stay the hell away from PHP (;))

On Thu, Apr 14, 2011 at 12:15 AM, AM <amcgeoch@gmail.com> wrote:

> If you look at the y scale (requests per second) against flask and bottle.
> Bottle drops over the three tests from around 3100 req/s to 1800 req/s and
> Flask drops from 2100 req/s to 1400 req/s.
>
> Over all Flask drops by 700 req/s over the three tests, whereas Bottle
> drops 1300 req/s over the three tests.
>
> If you go further. By test Flask drops 2100 > 1900 (200 req/s) from the
> hello world test to the templating test. Then 1900 > 1400 (500 req/s) from
> the templating test to the template/db test.
>
> wonder what his database code looks like or if it could be improved, im
> guessing it probably can.
>
> Bottle is a little more consistent doping 3100 > 2500 (600 req/s) then 2500
> > 1800 (700 req/s)
>
> overall bottle does drop pretty heavily as soon as you move from basic
> hello world apps but it doesnt look that obvious in the graphics he
> provides.
>
> The bigger frameworks don't do much better doping by the thousand each
> time.
>
> It would be nice if his graphics actually had a consistent scale.
>
>
> On 13 April 2011 19:40, dag.odenhall@gmail.com <dag.odenhall@gmail.com>wrote:
>
>> I would say that Flask values elegance of API and source code over raw
>> performance, when the gains in performance would be negligible. That's
>> not to say Flask is slow, though.
>>
>>
>> On 13 April 2011 20:53,  <mikecampo@gmail.com> wrote:
>> > Hey Jonathan,
>> >
>> > I think the whole point to Flask being a micro framework is so that
>> anyone
>> > can easily modify it to suit their needs. Out of the box it's fast and
>> lean.
>> > This works for most people and you won't notice any speed issues.
>> > If you need Flask to run a service like Twitter, you have quite a lot of
>> > work ahead of you. But any large web app will require substantial
>> deviation
>> > from the original framework, whether homebrew or not.
>> >
>> > Sent from my BlackBerry device on the Rogers Wireless Network
>> >
>> > ________________________________
>> > From: Jonathan Chen <tamasiaina@gmail.com>
>> > Sender: flask@librelist.com
>> > Date: Wed, 13 Apr 2011 11:32:36 -0700
>> > To: <flask@librelist.com>
>> > ReplyTo: flask@librelist.com
>> > Subject: [flask] Web Framework shootout
>> > http://blog.curiasolutions.com/the-great-web-framework-shootout/
>> >
>> > Has anybody read this article? It seems like bottle and flask do pretty
>> well
>> > compared to other frameworks when dealing with templates and a database.
>> Now
>> > I know that these articles are not really containing the full picture,
>> but
>> > is there any way we can try to make flask faster and optimize it?
>> >
>> >
>> > ~Jonathan C.
>> >
>>
>
>