Looking for technical reviewer for my upcoming Julia book


#1

I’m working with Packt on “Julia v1.0 By Example”, a beginner’s book about Julia. It covers various topics related to data science and general and web computing. I’m putting a lot of time, energy and love into the book - I’m hoping it’s going to help some people find their way into programming with Julia and enjoy it as much as I do. I really want to come out as a valuable resource and contribute to the growth of the language and its great community.

Part of it is that we need the support of a technical reviewer, and again, I’m reaching out to the awesome Julia users :slight_smile:

I’m looking for a partner to work with me on the project as the technical reviewer: go over the book, provide feedback about the code, ideas for improvement of the technical info, squash bugs, etc. If you’re interested, please message me, I’ll put you in touch with the Packt team for more details. Thanks!


#2

That’s good news!

Nothing to see with Genie.jl inside by the way?


#3

While in general I am very happy to see new books appear on Julia, and I also think that examples are the best way to do it, are you sure that this is the best time for it given the current state of transition in the DataFrames ecosystem? I suspect that by the time the ink dries, most examples will not run.


#4

It’s pretty much work in progress. The book targets the v1.0 release of Julia - with a preliminary publishing date for spring 2018. So yes, some of the APIs will change for sure and I’ll just have to update the code accordingly. But I expect that by the time we see a Julia v1.0 release, the DataFrames public APIs will be stable. What do you think? Also, can you please point me to the documentation regarding breaking API changes in the upcoming DataFrames?


#5

Probably not :slight_smile: The web development parts are more about supporting data science features (web mining, exposing data on the web, etc), so very light stuff. So far I’m going with HTTP.jl.

For Genie, the plan is to use my newly acquired literary skills to write the long overdue guides and proper docs. :blush:


#6

I am not the best person to ask about his, but this post could be a good starting point:


Some of the issues/PRs mentioned there have been resolved already, look on Github for new ones :wink:

Also
https://github.com/JuliaData/DataFrames.jl/issues/1232


#7

Thank’s so much @Tamas_Papp!


#8

There is no need to focus or even use dataframes. Scikit-learn is all build under the assumtion that the input and output data comes from arrays.


#9

The docs for the next DataFrames release are already online (there’s also an open PR to improve them): https://juliadata.github.io/DataFrames.jl/latest

Of course there’s no guaranty the API won’t change during the next year (though in general code won’t break without printing a deprecation first).


#10

Certainly, but

  1. a DataFrames-like representation with named columns, and
  2. consistent handling of missing data

make data analysis much more convenient. That’s why so many languages have these features.


#11

Data science without dataframes would feel like a step back. ScikitLearn.jl has support for them ported over from sklearn-pandas.


#12

Oh no! You’re going to finish your book first! Oh well…

I would absolutely love to help review your book. Perhaps you can help review mine (after I actually get around to writing it).


#13

Good news you will do this stuff for Genie.jl later Adrian.
I will be interested to read it.
May be i could help by drawing a few sketchs or technical diagrams.


#14

That would be much appreciated - I’m thinking for v1.0 I can start with the initial draft then we can use Git(Hub) to improve and maintain.


#15

I looked for an email address, but could not find one, so my response has to be public.

I have a lot of students in a class in Winter quarter. I would be happy to strongly suggest a book for learning julia, even if it is just a prototype. please email me: ivo.welch@gmail.com

/iaw


#16

Thanks Ivo - I wrote you a PM in Discourse to follow up. You might wanna edit your public message so that the email address isn’t harvested. :male_detective:


#17

your and my email have probably been harvested a gazillion times already, if only from emails we have sent to non-friends. the only thing that protects us are good spam filters, like gmail’s.

I am glad that julia has not closed down its syntax and semantics yet, but the drawback is that googling for solutions can often provide old, outdated, non-working answers. this is an obstacle to adoption. c’est la vis. for your book to succeed, keeping examples working all the way up to 1.0 is essential. (you should aggressively drop obsolete julia versions IMHO.) . I am looking forward to it. the more working and checked examples, the better.

regards,

/iaw


#18

Yeah, I guess I’m old-fashioned, still obfuscating email addresses on websites :slight_smile:

Absolutely - it’s not just Julia itself (which will hopefully evolve smoothly into v1.0) but (mostly) the packages. The previously discussed DataFrames, or WebServer and Requests morphing into HTTP.jl are just some of the key packages used in the examples. Staying up to date is an uphill battle now, but things will hopefully settle down around v1. Fingers crossed.


#19

Is that title set in stone? Somehow the inclusion of “v1.0” in the title strikes me as being too specific, especially considering that v1.0 will be the only officially supported version at publication time.


#20

Yes, good point. I’m not sure what’s the justification behind this very specific title, but I’ll ask.

The minor version is especially troubling since it suggests that it won’t work for a v1.1. Which it should, since minor versions are non-breaking.

The major version number is OK I think, as it implies breaking changes – and is a good opportunity to come up with a 2nd edition, for “v2”.