My story is similar.  I have no reservations recommending Julia to 
colleagues what-so-ever.  

I write optimization software for a living,  and I've used the basic parts 
of Matlab or Python + addons in past, but always got so annoyed that I 
switched back and forth between them for different projects.

With Julia I finally found something that I really like.  I can structure 
my Julia prototype in a way that's easily portable to C later on, and it's 
a bliss to write larger projects. 

I've written Julia prototypes for a complete conic interior-point 
optimizer, a graph partition algorithm for reordering large sparse 
matrices, and a toolbox for polynomial optimization using semidefinite 
programming, and the code is nice and clean and scales well.  I could not 
have done that with either Matlab or Python.

Things have broken for me in Julia when I upgrade,  but I've always been 
able to update my projects in a couple of hours (after line-number 
reporting in error messages has become sensible).



Den fredag den 6. marts 2015 kl. 09.42.29 UTC+1 skrev Daniel Carrera:
>
> On Thursday, 5 March 2015 18:49:24 UTC+1, Stefan Karpinski wrote:
>>
>> It's the people who are desperately unhappy with what they currently use 
>> that might really benefit – and those people do exist. 
>>
>
> *raises his hand*
>
> That is exactly me. For years I have wanted a language for scientific 
> computing with a nice syntax, nice API, and open source. I never managed to 
> like Python + NumPy. I was using Octave, and I was on the mailing list 
> asking if they'd consider making a few more improvements on the Matlab 
> syntax when someone said "have a look at Julia, it has the things you are 
> asking for". That was about a year after the official release of Julia (I 
> think).
>
>  
>
>> but you'll have to deal with sometimes implementing things that other 
>> language already have packages for and with packages sometimes breaking 
>> when you upgrade them (the secret is don't upgrade often).
>>
>
> Yeah. When my last paper was getting ready to be submitted I upgraded 
> *nothing*, even though a new and faster version of Julia had just come out. 
> I could not take the risk of some of my scripts breaking at the last minute.
>
>  
>
>> As long as that tradeoff is clear, I think it's ok to recommend Julia, 
>> but one does have to set expectations honestly and not oversell it.
>>
>
> Yeah. A couple of people have asked me about Julia and I tell them 
> something similar to what you just wrote.
>
> Cheers,
> Daniel.
>  
>

Reply via email to