Category:Phix

Phix
This programming language may be used to instruct a computer to perform a task.

Official website

Execution method:	Interpreted or compiled
Garbage collected:	Yes
Parameter passing methods:	By reference, By value
Type expression:	Explicit
Type checking:	Dynamic, Static
Lang tag(s):	phix, Phix
See Also:	BNF Grammar for Phix Phix compared to other languages

Listed below are all of the tasks on Rosetta Code which have been solved using Phix.

Phix is a self-hosted hybrid interpreter/compiler, developed by Pete Lomax. It is very easy to use, and similar to Euphoria.

On 26th December 2021, Phix became the first programming language on this site to have submissions for 100% of all tasks.
(On 20th June 2022 Wren became the second, albeit with 13+66 of those embedded that I, for one, have been unable to run.)

A simple 30MB download contains a pre-compiled executable, all the sources, and everything needed to recompile them, in under 20 seconds. The download also contains a full-featured programmer's editor and about 1000 demo programs.

You can also run Phix on replit here (1.0.3 Lnx 64 bit), and several manually transpiled and uploaded entries here

Perhaps the most striking feature of Phix is that it has just five builtin data types:

        <-------- object --------->
        |                |
        +-atom           +-sequence
          |                |
          +-integer        +-string

Despite such apparent simplicity, or perhaps precisely because of it, Phix programs are pretty fast - not quite achieving the runtime performance of C or assembly, but making up for it with a very fast edit/run cycle and proper human-readable messages should anything go wrong (even in shipped pre-compiled executables). Sequences are the real powerhouse of Phix. The one type covers lists, queues, tables, trees, and arrays, with strings being the subset that is array of character. They can grow and shrink automatically without any memory management overhead. For example if s="food" then s[2..3]="e" makes s "fed", and then s[2..1]="east" makes s "feasted".

Phix applies the principle of least surprise, for instance in some languages myproc(list) or res = myfunc(list) can mangle list, whereas in Phix if you actually want that to happen you would code list = myproc(list) (and myproc would need to become a function) or {res,list} = myfunc(list). Likewise 1/2 is 0.5 (not 0, unless you explicitly ask for the floor()) and 0-1 is -1 (not +MAXINT). A core tenet is that for any line of code there is one and only one possible interpretation of it, and said meaning is utterly intuitive. True fact: the given answer for the question "describe what f(a++) does" in "More Exceptional C++" lists 4 possibilities for f, and 3 for a (so 12 in total) and uses the phrase "could mean just about anything" not once but twice - shudder.

Phix supports object orientation, but does not enforce it, and can achieve many of the claimed benefits in a much simpler way. In fact I have been looking for proof that object orientation actually improves productivity compared to other paradigms for decades, and never found it. One other thing I have never found is a "good object orientated design", and reached the conclusion that mythical creature simply does not exist, at least not as a separate entity as opposed to some ethereal quality of the finished code. Feel free to argue that one on User_talk:Petelomax. Phix offers perfectly decent encapsulation at the file level, proving that is not the sole purview of oo, implements polymorphism far more elegantly than C-based languages and far safer than duck-typed languages, and as for the third pillar of oop, inheritance, well isn't the current mantra "favour composition over inheritance"?

Technically Phix does not have a garbage collector, instead it is reference counted, so memory is automatically reclaimed whenever a reference count drops to zero, and hence there are no sudden pauses, although the reference counting can itself reduce performance.

The fundamental goal of Phix is to make debugging easier, a whole subject area that does not seem to get the attention it deserves, despite several studies finding that most programmers actually spend between 70 and 90% of their time debugging. Most of us tend to think that getting just-written code to work is somehow not debugging, but that clipboard-toting researcher stood behind you would disagree.

Since 1.0.0 you can also run (many/most) Phix programs in a browser, by using pwa/p2js (as included in the distro) to transpile the program to JavaScript (plus a bit of standard HTML and CSS). If you see "with javascript_semantics" it means the program has been tested and works fine, both on the (Windows/Linux) desktop/Phix and in a browser via pwa/p2js, and in fact unix_dict() has been added to avoid reading unixdict.txt. Conversely, "without js" being present on a rosettacode entry means that (for instance) it uses the local file system, and is therefore desktop/Phix only. It is worth pointing out that "with js" and deep_copy() have proved staggeringly effective and nowhere near as painful as first feared. In just a few months (1.0.0 was released in July 2021) I have tested, fixed where necessary, and marked as javascript compatible over 1,200 rosettacode entries for Phix, and only one seventh of that have as yet been marked incompatible, despite things like file i/o being prohibited within a web browser. The Phix/online subcategory currently lists 99 entries deemed interesting enough to bother (manually) uploading to the Phix website (ping me should any stop working), and the next goal is to get pwa/p2js itself runnning online, but obviously without using file i/o directly. Another long term goal is to get Phix running on ARM, and since that would be utterly pointless without a viable GUI, I am currently elbow deep in writing a new cross-platform GUI, as a stepping stone (for both ARM and p2js/online).

See also

• Phix site
• github
• Tasks_not_implemented_in_Phix