Re: Release vibe.d 0.7.27

[sigod via Digitalmars-d-announce]

On Sunday, 14 February 2016 at 08:17:34 UTC, Sönke Ludwig wrote:

Am 11.02.2016 um 00:24 schrieb sigod:

Did some benchmarks between `std.net.curl.get` and
`vibe.http.client.requestHTTP`. Only GET requests.

100 requests, ~1.4mb file:

 curl total: 131304, average: 1 sec and 313 ms
 vibe total:  21975, average: 219 ms

52 different files:

 curl total: 24851, average: 477 ms
 vibe total: 11290, average: 217 ms

50 different files (excluded 2 of the biggest ones):

 curl total: 20892, average: 417 ms
 vibe total: 11368, average: 227 ms

(Looks like `std.net.curl.get` doesn't like if file is bigger 
than ~1mb.)


Is vibe.d's API really that fast? Or am I doing something 
wrong?


How fast was the network connection in that case? Could it make 
a difference if keep-alive connections are used or not? Were 
the requests done in parallel or in sequence? I certainly 
wouldn't expect curl to be slower for a simple sequential 
transfer of a single file, but who knows.


https://gist.github.com/sigod/c78c61ac6118fa9fda26

I'm getting something like this:

HTTPS:
curl total: 23401, average: 458ms
vibe total: 12136, average: 237ms
HTTP:
curl total: 5577, average: 278ms
vibe total: 4268, average: 213ms

Windows 7 x86, dmd 2.070.0

Re: Release vibe.d 0.7.27

[Sönke Ludwig via Digitalmars-d-announce]

Am 11.02.2016 um 00:24 schrieb sigod:

Did some benchmarks between `std.net.curl.get` and
`vibe.http.client.requestHTTP`. Only GET requests.

100 requests, ~1.4mb file:

 curl total: 131304, average: 1 sec and 313 ms
 vibe total:  21975, average: 219 ms

52 different files:

 curl total: 24851, average: 477 ms
 vibe total: 11290, average: 217 ms

50 different files (excluded 2 of the biggest ones):

 curl total: 20892, average: 417 ms
 vibe total: 11368, average: 227 ms

(Looks like `std.net.curl.get` doesn't like if file is bigger than ~1mb.)

Is vibe.d's API really that fast? Or am I doing something wrong?


How fast was the network connection in that case? Could it make a 
difference if keep-alive connections are used or not? Were the requests 
done in parallel or in sequence? I certainly wouldn't expect curl to be 
slower for a simple sequential transfer of a single file, but who knows.

Re: Release vibe.d 0.7.27

[Daniel Kozak via Digitalmars-d-announce]

Dne 10.2.2016 v 09:50 Iain Buclaw via Digitalmars-d-announce napsal(a):
On 9 February 2016 at 20:16, Sönke Ludwig 
> wrote:


This release brings some larger changes:

 - The library has been split up into sub packages: code, utils, data,
   http, mail, diet, mongodb, redis and web. This is an intermediate
   step to moving the individual packages out to separate repositories
   with independent version numbers.

 - A lot of work went into performance tuning. Single-core performance
   of the HTTP server is improved by about +50% and multi-core
   performance scales properly again after excessive lock contention
   sneaked in in one of the previous releases. The number of worker
   threads is now also properly determined on all systems (including
   multi-CPU), which should fix the numbers for multi-threaded
   benchmarks (an update to the TechEmpower benchmark suite is on the
   way).

 - The REST interface generator now supports modelling collections
with
   native D syntax using Collection!T. It also adds support for CORS.

 - The std.concurrency integration has been fixed and re-enabled - you
   can now use std.concurrency without worrying about blocking the
   event loop. In case of problems (std.concurrency doesn't support
   passing certain kinds of values), the old implementation can still
   be accessed as sendCompat/receiveCompat/...

 - Compiles on 2.066.0 up to 2.070.0. Note that this will be the last
   release that supports the 2.066.x frontend. The next release will
   require at least 2.067.0 or maybe even 2.068.0 (still TBD).
This may
   unfortunately rule out GDC for the time being.


Not a problem.  We may be on 2.068 sooner than you think (I know that 
2.067 has been in branch/PR forever ;-)


Iain.

Wow, that would be awesome :)

Re: Release vibe.d 0.7.27

[Sönke Ludwig via Digitalmars-d-announce]

Am 10.02.2016 um 09:50 schrieb Iain Buclaw via Digitalmars-d-announce:


Not a problem.  We may be on 2.068 sooner than you think (I know that
2.067 has been in branch/PR forever ;-)

Iain.


Really glad to hear that! BTW, is anything moving w.r.t. integrating GDC 
(and LDC) to the auto tester for the DMD frontend?

Re: Release vibe.d 0.7.27

[Iain Buclaw via Digitalmars-d-announce]

On 10 February 2016 at 10:49, Sönke Ludwig <
digitalmars-d-announce@puremagic.com> wrote:

> Am 10.02.2016 um 09:50 schrieb Iain Buclaw via Digitalmars-d-announce:
>
>>
>> Not a problem.  We may be on 2.068 sooner than you think (I know that
>> 2.067 has been in branch/PR forever ;-)
>>
>> Iain.
>>
>
> Really glad to hear that! BTW, is anything moving w.r.t. integrating GDC
> (and LDC) to the auto tester for the DMD frontend?
>

Don't think so.  We have Travis CI building DMD with GDC and LDC for each
PR, however I don't know whether or not the auto-tester takes the
success/failure of these builds into account.

Re: Release vibe.d 0.7.27

[Robert M. Münch via Digitalmars-d-announce]

On 2016-02-09 19:16:49 +, Snke Ludwig said:


  - Full list of changes: http://vibed.org/blog/posts/vibe-release-0.7.27


Great stuff!! Thanks a lot for your engagement.

--
Robert M. Münch
http://www.saphirion.com
smarter | better | faster

Re: Release vibe.d 0.7.27

[Iain Buclaw via Digitalmars-d-announce]

On 9 February 2016 at 20:16, Sönke Ludwig <
digitalmars-d-announce@puremagic.com> wrote:

> This release brings some larger changes:
>
>  - The library has been split up into sub packages: code, utils, data,
>http, mail, diet, mongodb, redis and web. This is an intermediate
>step to moving the individual packages out to separate repositories
>with independent version numbers.
>
>  - A lot of work went into performance tuning. Single-core performance
>of the HTTP server is improved by about +50% and multi-core
>performance scales properly again after excessive lock contention
>sneaked in in one of the previous releases. The number of worker
>threads is now also properly determined on all systems (including
>multi-CPU), which should fix the numbers for multi-threaded
>benchmarks (an update to the TechEmpower benchmark suite is on the
>way).
>
>  - The REST interface generator now supports modelling collections with
>native D syntax using Collection!T. It also adds support for CORS.
>
>  - The std.concurrency integration has been fixed and re-enabled - you
>can now use std.concurrency without worrying about blocking the
>event loop. In case of problems (std.concurrency doesn't support
>passing certain kinds of values), the old implementation can still
>be accessed as sendCompat/receiveCompat/...
>
>  - Compiles on 2.066.0 up to 2.070.0. Note that this will be the last
>release that supports the 2.066.x frontend. The next release will
>require at least 2.067.0 or maybe even 2.068.0 (still TBD). This may
>unfortunately rule out GDC for the time being.
>
>
Not a problem.  We may be on 2.068 sooner than you think (I know that 2.067
has been in branch/PR forever ;-)

Iain.

Re: Release vibe.d 0.7.27

[Suliman via Digitalmars-d-announce]

On Wednesday, 10 February 2016 at 08:50:54 UTC, Iain Buclaw wrote:
On 9 February 2016 at 20:16, Sönke Ludwig < 
digitalmars-d-announce@puremagic.com> wrote:



This release brings some larger changes:

 - The library has been split up into sub packages: code, 
utils, data,
   http, mail, diet, mongodb, redis and web. This is an 
intermediate
   step to moving the individual packages out to separate 
repositories

   with independent version numbers.

 - A lot of work went into performance tuning. Single-core 
performance

   of the HTTP server is improved by about +50% and multi-core
   performance scales properly again after excessive lock 
contention
   sneaked in in one of the previous releases. The number of 
worker
   threads is now also properly determined on all systems 
(including

   multi-CPU), which should fix the numbers for multi-threaded
   benchmarks (an update to the TechEmpower benchmark suite is 
on the

   way).

 - The REST interface generator now supports modelling 
collections with
   native D syntax using Collection!T. It also adds support 
for CORS.


 - The std.concurrency integration has been fixed and 
re-enabled - you
   can now use std.concurrency without worrying about blocking 
the
   event loop. In case of problems (std.concurrency doesn't 
support
   passing certain kinds of values), the old implementation 
can still

   be accessed as sendCompat/receiveCompat/...

 - Compiles on 2.066.0 up to 2.070.0. Note that this will be 
the last
   release that supports the 2.066.x frontend. The next 
release will
   require at least 2.067.0 or maybe even 2.068.0 (still TBD). 
This may

   unfortunately rule out GDC for the time being.


Not a problem.  We may be on 2.068 sooner than you think (I 
know that 2.067 has been in branch/PR forever ;-)


Iain.


Does your company use D?

Re: Release vibe.d 0.7.27

[sigod via Digitalmars-d-announce]

Did some benchmarks between `std.net.curl.get` and 
`vibe.http.client.requestHTTP`. Only GET requests.


100 requests, ~1.4mb file:

curl total: 131304, average: 1 sec and 313 ms
vibe total:  21975, average: 219 ms

52 different files:

curl total: 24851, average: 477 ms
vibe total: 11290, average: 217 ms

50 different files (excluded 2 of the biggest ones):

curl total: 20892, average: 417 ms
vibe total: 11368, average: 227 ms

(Looks like `std.net.curl.get` doesn't like if file is bigger 
than ~1mb.)


Is vibe.d's API really that fast? Or am I doing something wrong?

Re: Release vibe.d 0.7.27

[extrawurst via Digitalmars-d-announce]

On Tuesday, 9 February 2016 at 19:16:49 UTC, Sönke Ludwig wrote:

This release brings some larger changes:

 - The library has been split up into sub packages: code, 
utils, data,
   http, mail, diet, mongodb, redis and web. This is an 
intermediate
   step to moving the individual packages out to separate 
repositories

   with independent version numbers.

[...]


Awesome! Thanks for all the hard work.

--Stephan

Release vibe.d 0.7.27

[Sönke Ludwig via Digitalmars-d-announce]

This release brings some larger changes:

 - The library has been split up into sub packages: code, utils, data,
   http, mail, diet, mongodb, redis and web. This is an intermediate
   step to moving the individual packages out to separate repositories
   with independent version numbers.

 - A lot of work went into performance tuning. Single-core performance
   of the HTTP server is improved by about +50% and multi-core
   performance scales properly again after excessive lock contention
   sneaked in in one of the previous releases. The number of worker
   threads is now also properly determined on all systems (including
   multi-CPU), which should fix the numbers for multi-threaded
   benchmarks (an update to the TechEmpower benchmark suite is on the
   way).

 - The REST interface generator now supports modelling collections with
   native D syntax using Collection!T. It also adds support for CORS.

 - The std.concurrency integration has been fixed and re-enabled - you
   can now use std.concurrency without worrying about blocking the
   event loop. In case of problems (std.concurrency doesn't support
   passing certain kinds of values), the old implementation can still
   be accessed as sendCompat/receiveCompat/...

 - Compiles on 2.066.0 up to 2.070.0. Note that this will be the last
   release that supports the 2.066.x frontend. The next release will
   require at least 2.067.0 or maybe even 2.068.0 (still TBD). This may
   unfortunately rule out GDC for the time being.

 - Full list of changes: http://vibed.org/blog/posts/vibe-release-0.7.27

Homepage: http://vibed.org/
DUB package: http://code.dlang.org/packages/vibe-d
GitHub: https://github.com/rejectedsoftware/vibe.d

Re: Release vibe.d 0.7.27

[Nick B via Digitalmars-d-announce]

On Tuesday, 9 February 2016 at 19:16:49 UTC, Sönke Ludwig wrote:

This release brings some larger changes:

 - A lot of work went into performance tuning. Single-core 
performance

   of the HTTP server is improved by about +50% and multi-core
   performance scales properly again after excessive lock 
contention
   sneaked in in one of the previous releases. The number of 
worker
   threads is now also properly determined on all systems 
(including

   multi-CPU), which should fix the numbers for multi-threaded
   benchmarks (an update to the TechEmpower benchmark suite is 
on the

   way).

I look forward to reading the numbers, and seeing how it 
compares, to other web servers :)


Nick

Re: Release vibe.d 0.7.27

[Dicebot via Digitalmars-d-announce]

On 02/09/2016 09:16 PM, Sönke Ludwig wrote:
> This release brings some larger changes:
> 
> ...

Changelog looks very exciting, good work!

Re: Release vibe.d 0.7.27

[Andrei Alexandrescu via Digitalmars-d-announce]

On 2/9/16 2:16 PM, Sönke Ludwig wrote:

  - Full list of changes: http://vibed.org/blog/posts/vibe-release-0.7.27

Homepage: http://vibed.org/
DUB package: http://code.dlang.org/packages/vibe-d
GitHub: https://github.com/rejectedsoftware/vibe.d


Congrats!! -- Andrei