Hi,
I am running Julia on a single machine with multiple processes. I am
porting code to work on a cluster with a shared filesystem.
The statement:
@everywhere include("/path/to/file")
works just fine.
The statement
@everywhere include(fname)
does not.
Here's the error message:
e
> There's an open pull request on the website repository -
> https://github.com/JuliaLang/julialang.github.com/pull/132
To clarify, the Julia 0.3 numbers are attached to the PR and there is
nothing wrong with them. However, there _are_ issues with the C, Fortran,
Java, Octave, and R microbenchmark
On Thursday, October 23, 2014 9:13:06 AM UTC-4, Tony Fong wrote:
>
> I also have a similar need. If I want to be more fancy, what may be the
> strategy to address the following 2-way communication?
> * UI -> worker: send pause, abort
> * worker -> UI: progress report, summary state (presumably t
No idea, but a thought: wouldn't be too bad to have a @magic macro applied
to the for loop - would even be kind of a warning to the reader that macro
shenanigans are coming up.
On Thursday, October 23, 2014 5:18:27 PM UTC-4, Simon Byrne wrote:
>
> Is it possible to determine whether or not it is
Is it possible to determine whether or not it is the first pass of a loop
entirely from within the loop, i.e. without adding extra statements outside
the loop?
The context is that I've been thinking about how to avoid reallocating
arrays on each loop, for instance something like this will reall
Understandably. But let's not repeat the mistake that clang made and leave
up benchmarks against old versions of everything much longer than they
should be there. They should either be updated - say once a year? - or
removed.
On Thursday, October 23, 2014 1:11:48 PM UTC-7, John Myles White wro
FWIW, I think some of us are pretty burnt out with benchmarks and the politics
involved with their "validity".
-- John
> On Oct 23, 2014, at 12:57 PM, Tony Kelman wrote:
>
> There's an open pull request on the website repository -
> https://github.com/JuliaLang/julialang.github.com/pull/132
Whoops, clicked post before finishing that thought process.
Can anyone think of some way of isolating ourselves from these breakages?
Anyone familiar with how RPM mirroring works? Would it be feasible to set
up our own mirror, and configure it in a way that runs an automatic sanity
check every
Oh. Looks like there's a major problem
upstream: https://build.opensuse.org/project/monitor/windows:mingw:win64
Hopefully they can fix it?
On Thursday, October 23, 2014 12:53:53 PM UTC-7, Isaiah wrote:
>
> That is surprising. I just tested WinRPM on 32-bit two days ago for 0.3.2
>
> On Thu, Oct
There's an open pull request on the website repository
- https://github.com/JuliaLang/julialang.github.com/pull/132
There are quite a few unresolved issues/questions about the benchmarks,
which it looks like no one has put much effort into addressing over the
past few months.
On Thursday, Oct
That is surprising. I just tested WinRPM on 32-bit two days ago for 0.3.2
On Thu, Oct 23, 2014 at 3:45 PM, Tony Kelman wrote:
> Yeah, something is definitely wrong with the repo metadata again. Not sure
> what yet.
>
> > (I've already tried with the 32 bit but that goes wrong even earlier in
> t
Yeah, something is definitely wrong with the repo metadata again. Not sure
what yet.
> (I've already tried with the 32 bit but that goes wrong even earlier in
the process -- WinRPM doesn't install).
Can you please elaborate on that? Exactly what goes wrong? WinRPM not
installing on 32 bit is a
It seems it was mostly ArrayViews (see other thread) and DataStructures,
although that may be masking other issues.
On Thursday, October 23, 2014 3:04:32 PM UTC-4, Iain Dunning wrote:
>
> Well, maybe not everything, but a lot of packages:
> http://pkg.julialang.org/pulse.html
>
> I just uploaded
Well, maybe not everything, but a lot of packages:
http://pkg.julialang.org/pulse.html
I just uploaded the testing results now, and I'm triaging them to determine
the root packages causing other packages to fail (really need to automate
that...)
Friendly reminder to all that Things Break on Ju
I've been asking myself the same as Uwe. Is there any place where we can
find this information?
Greetings,
Pablo
On Thursday, October 23, 2014 10:50:19 AM Douglas Bates wrote:
> What is the state of "Array Nirvana" in 0.4? I am happy to avoid using
> ArrayViews if I can get contiguous views in 0.4
My personal plan is to fix #8504 and then merge #8501. Then look into
incorporating contiguous views, reshape
I saw this today too.
-viral
On Thursday, October 23, 2014 11:20:19 PM UTC+5:30, Douglas Bates wrote:
>
> I think this is due to a recent change in 0.4.0. The declaration
>
> # use ContiguousView when contiguousness can be determined statically
> immutable ContiguousView{T,N,Arr<:Array{T}} <: Ar
The T parameter in the type signature of Array{T} below was recently
changed from being silently ignored to not being parsed. I think Jeff may
have merged this in his call_overload branch?
On Thursday, October 23, 2014, Douglas Bates wrote:
> I think this is due to a recent change in 0.4.0. The
None that I know of. It may help to file an issue against the Distributions
package - but the question of translations and keeping them in sync is a
larger issue.
-viral
On Thursday, October 23, 2014 9:31:28 PM UTC+5:30, Carlos Lesmes wrote:
>
> Hi,
> I would like to translate to spanish the d
I enjoyed it too while reviewing. In fact I just met up with Bruce Tate a
couple of weeks back here in Bangalore, and it was fun talking about Julia
and other stuff. His story about how they needed a seventh language and how
it ended up being Julia is really entertaining and one should ask him..
I think this is due to a recent change in 0.4.0. The declaration
# use ContiguousView when contiguousness can be determined statically
immutable ContiguousView{T,N,Arr<:Array{T}} <: ArrayView{T,N,N}
arr::Arr
offset::Int
len::Int
shp::NTuple{N,Int}
end
throws an ERROR: T not defin
Hello,
on the main Julia website there are still the benchmark results of Julia
0.2 published.
Did anybody try to run these benchmarks with Julia 0.3x ?
How does Julia 0.3 compare to Julia 0.2 and to the other programming
languages?
Best regards:
Uwe
I got to review the Julia chapter and really enjoyed it. It's quite good.
Looking forward to reading the whole book! Please to submit a PR adding a link
to the book.
> On Oct 23, 2014, at 12:02 PM, John Myles White
> wrote:
>
> This is a good idea. Please submit a pull request that adds link
This is a good idea. Please submit a pull request that adds link to that
section.
-- John
On Oct 23, 2014, at 6:05 AM, Ivo Balbaert wrote:
> Hi,
>
> Within some weeks the Pragmatic Programmers will publish the book: "Seven
> More Languages in Seven Weeks"
> by Bruce Tate, Fred Daoud, Jack
Hi,
I would like to translate to spanish the distributions package docs.
Are there any guidelines to do it in Github?
CL
I have a plot where I've got enough points packed together that some
are indistinguishable. What's a good way to fix this? With ggplot2 I'd
either make the points translucent or add jitter, but I can't figure
out how do to either with Gadfly.
I'm plotting with commands like
plot(ndf[ndf[:level].==
Completely devectorisation can sometimes be useful, but I have come to
prefer writing linear algebra code close to the style of LAPACK where you
organise the calculations in BLAS calls. Not only for speed, but also for
making the code easier to read and infer the structure of the algorithm.
Med ve
>
> On Thursday, 23 October 2014 15:02:23 UTC+1, Tim Holy wrote:
>>
>> I believe I mentioned A_mul_B! and friends. For the subtraction, check
>> out
>> NumericExtensions, specifically the subtract! function.
>
>
> Also I would like to shamelessly plug InplaceOps:
> https://github.com/simonbyrne
Dňa štvrtok, 23. októbra 2014 16:21:11 UTC+2 Andreas Noack napísal(-a):
>
> I missed that you were already using the ArrayViews package. Sorry for
> that.
>
> The way you have written the rank one update in line 14 is very expensive.
> Unfortunately, we are not making all of BLAS available throu
Personally, I greatly appreciate the shameless plug. There are too many cool
things available to keep track of them all, so it's great when people
advertise their own code.
I'll have to try this, it looks quite handy.
--Tim
On Thursday, October 23, 2014 07:31:34 AM Simon Byrne wrote:
> On Thur
Dňa štvrtok, 23. októbra 2014 16:02:23 UTC+2 Tim Holy napísal(-a):
>
> I believe I mentioned A_mul_B! and friends. For the subtraction, check out
> NumericExtensions, specifically the subtract! function.
>
> --Tim
>
> Yes, you did. Exploration along this line is in progress. :) I will also
chec
On Thursday, 23 October 2014 15:02:23 UTC+1, Tim Holy wrote:
>
> I believe I mentioned A_mul_B! and friends. For the subtraction, check out
> NumericExtensions, specifically the subtract! function.
Also I would like to shamelessly plug InplaceOps:
https://github.com/simonbyrne/InplaceOps.jl
wh
I missed that you were already using the ArrayViews package. Sorry for
that.
The way you have written the rank one update in line 14 is very expensive.
Unfortunately, we are not making all of BLAS available through our generic
mat-vec interface right now, but the whole of BLAS is wrapped in the BL
Hello,
I have some trouble installing Winston on Windows. To make this as clean as
possible I uninstalled Julia, then deleted my $HOME/.julia directory, and
reinstalled Julia 64 bit (I've already tried with the 32 bit but that goes
wrong even earlier in the process -- WinRPM doesn't install).
I believe I mentioned A_mul_B! and friends. For the subtraction, check out
NumericExtensions, specifically the subtract! function.
--Tim
On Thursday, October 23, 2014 06:43:39 AM Ján Dolinský wrote:
> > I'd like to approach this speed at the end.
> >
> >
> > I don't think it is possible in Jul
You also might check out the Devectorize.jl package.
Cheers,
Kevin
On Thursday, October 23, 2014, Tim Holy wrote:
> For some of your cases, copy!(t, a) or fill!(t, 1.0+2.0) may be what you
> want.
> --Tim
>
> On Thursday, October 23, 2014 01:58:48 AM accouju...@gmail.com
> wrote:
> > > Thos
>
> I'd like to approach this speed at the end.
>
>
> I don't think it is possible in Julia right now without using dirty tricks
> such as passing pointers around. You'd like to get the speed from BLAS by
> operating on panels of your matrix, but you'd like to avoid the copying and
> reallocati
I also have a similar need. If I want to be more fancy, what may be the
strategy to address the following 2-way communication?
* UI -> worker: send pause, abort
* worker -> UI: progress report, summary state (presumably to allow UI to
do fancier presentation)
I'm thinking about looking into TcpS
Hi,
Within some weeks the Pragmatic Programmers will publish the book: "Seven
More Languages in Seven Weeks"
by Bruce Tate, Fred Daoud, Jack Moffitt, Ian Dees:
https://pragprog.com/book/7lang/seven-more-languages-in-seven-weeks
It contains a tutorial Chapter for programmers new to Julia, whi
A new version of ApproxFun.jl has been tagged, with several major changes
behind the scenes. It's now possible to solve some PDEs on a disk, for
example the following solves Poisson equation Δ u = f with Dirichlet
boundary conditions:
d=Disk()
f=Fun((x,y)->exp(-10(x+.2).^2-20(y-
For some of your cases, copy!(t, a) or fill!(t, 1.0+2.0) may be what you want.
--Tim
On Thursday, October 23, 2014 01:58:48 AM accouju...@gmail.com wrote:
> > Those coming to Julia from languages like Python, R, Matlab/Octave, etc.
> > have it ingrained in them to "fear the loop" and to perform lo
On Thursday, October 23, 2014 12:43:57 AM accouju...@gmail.com wrote:
> > accoujulia, see the "performance tips" section of the manual, which
> > describes
> > several tools.
> Please bear in mind that I was using Julia 0.3 so I only looked at the 0.3
> documentation as I'm supposed to.
You are s
It’s worth noting that the way Julia parses expressions is open to
exploration if you know what functions to use.
julia> ex = :(-1^2) # quote the piece of code you'd like to know more about
:(-(1^2))
julia> Base.Meta.show_sexpr(ex) # a lisp-y representation of the expression
(:call, :-, (:call
Here are the operator precedence rules:
http://docs.julialang.org/en/release-0.3/manual/mathematical-operations/?highlight=operators#operator-precedence
so it's pretty clear that -1^2==-1
But for sure, it can be a bit confusing how the `-` is parsed:
julia> -1*-1
1
is probably parsed as:
julia
>
> Those coming to Julia from languages like Python, R, Matlab/Octave, etc.
> have it ingrained in them to "fear the loop" and to perform loop-like
> calculations in "verbose and convoluted" ways.
>
Please compare my Python version with yours. If I look at the Python
version in 2 month, I'll i
As John said, this is not about handling the exponential, but about
handling the minus sign.
But this isn't some sort of Julia quirk:
MATLAB:
>> -1^2
ans =
-1
Python:
>> -1**2
-1
R:
> -1^2[1] -1
Seems pretty consistent to me!
>
> Yes, check the link Tim Holy sent - its a command line option.
>
His e-mail has been sent 2 minutes before mine and I think I missed GG UI
notification for a new e-mail.
>
> accoujulia, see the "performance tips" section of the manual, which
> describes
> several tools.
>
> --Tim
>
Please bear in mind that I was using Julia 0.3 so I only looked at the 0.3
documentation as I'm supposed to. It's true that I should have thought
about putting @allocated in the
48 matches
Mail list logo