_{Jamie Balfour's}Personal blog

ZPE 1.12.9 is one of the most significant updates to ZPE since it started. It completes the major project of making all internal data types as instances of ZPEType, rather than relying on lower-level and abstract data types. This update is so significant because it now means higher performance due to less casting, better safety and ensuring that non-ZPEType values can no longer be injected into the program. 

Further to this, new syntax features include the use of colon terminated conditions and for parameters:

while $x < 100:
  print($x)
end while

Tuples are another feature of ZPE 1.12.9; they are different from class instances and objects entirely in that they are defined once but follow such a structure throughout their lifetime:

$x = (string, number, number)
$x = ("Hello world", 13, 7)

The new REQUIRE_EXPLICIT_VARIABLE_DECLARATION property and the requires_explicit_declaration directive both enforce that the user defines variables before they are used. This is a significant feature, which might become the default option in ZPE later on.

Finally, ZPE's editor has undergone a bit of overhaul. The icon is scaled down to look much smarter and suitable in internal windows and alert popups, whilst on macOS the logo used around the application is the logo found on the Dock. The scrollbars have been enhanced, too. And toggling dark mode now retains the position you were at when you toggled it. The final significant thing is that the console now runs in a separate process and can be stopped in between running a program, meaning if an infinite loop starts in the code, it will not affect the main editor. This will also come to my SQARL editor, which has also received a considerable number of improvements, including an icon at last.

ZPE 1.12.9 is now available to download from the usual place.

For an extremely long time, ZPE has relied on native Object types for return values. I never wanted to keep this because it meant a lot of type casting, type inference, potential bugs, a lack of security regarding what was passed around ZPE, and also a lot of type checking. This slowed down ZPE and meant it didn't feel particularly stable. 

As of version 1.12.9, ZPE is now fully locked down to the ZPEType return type, which includes ZPEStrings, ZPENumbers, ZPEBooleans, ZPEObjects, ZPEFunctions, ZPERecords, ZPETuples, ZPELists, ZPEArrays, ZPEMaps, and ZPEOrderedMaps. This reduces ZPE's flexibility, making the Core much more stable. 

As I say, this is one of the biggest updates to ZPE in a long time and it paves the way to further updates that will strengthen this.

What's really cool about this update, nothing has changed in the compiler so no scripts that have been compiled need to be changed, and in fact, nothing needs to be changed in any existing scripts.

ZPE 1.12.9 features a few major changes and improvements, but one stands out more than the others: the introduction of explicit variable declaration.

Explicit variable declaration means that variables must be declared before they are used, either in an assignment or for anything else. For example, if a variable $x isn't defined first, the compiler will detect this and throw an error. 

$x = 15

will throw an error, whereas

number $x = 0
$x = 15

will not.

In addition, ZPE gets closer to having all underlying data types as ZPETypes. This will enforce the fact that things like plugins can only return types that ZPE itself controls, which will lead to better type safety. As of ZPE 1.12.9, we now have the ZPENumber data type underneath. It took me a whopping 8 hours to bring it to ZPE, and the implementation isn't complete yet. Further, I aim to improve the way the ZPEString is managed so that fewer checks need to be done with them as well.

ZPE 1.12.7 brought many of the biggest updates to ZPE in a long time including bundling a YASS-based standard library that is compiled on start up, changing the way imports work and breakpoints.

Now ZPE 1.12.8 is just about ready for launch and it brings many big new features and changes.

The new features of ZPE 1.12.8 are:

• Namespaces have finally been added to ZPE
• Much better error handling
• The new -h bytes will list all byte codes available in ZPE and their representation, great for development
• New Javadoc based documentation is being added, a considerable amount has already been added to ZPE 1.12.8.
• Return types on functions can now be stated with -> as well as :
• Several new exceptions have been added to ZPE (this version is known as the 'exception version' as a result)
• Short anonymous functions () => {} now support typed parameters.
• Static functions and static function calls are now available too (they will soon support namespaces too)

The big changes in ZPE 1.12.8 are:

• The cache values file is now stored in the ZPE folder, rather than where ZPE is executed. It makes much more sense this way.
• ZPE introduced a new ZPERandomizer engine in July
• Log files have been changed so they no longer take up too much space on disk and they rotate more efficiently
• ZPE Online files are now compiled when they download to ZPE
• The JSON parser now supports the null value
• Regexps have been improved to reduce the time they require to compile and test
• The main introduction is more detailed than ever before, making it easier to find out about a ZAC than before
• std_out prints to the same line
• match statements now only support one else statement and it must be the last condition of the statement

There have been several bug fixes too:

• ZPE no longer keeps asking to install the standard library if you said no
• Shell return values are now ZPEStrings making them comply with the ZPE data type list
• The --man option stated that all functions return an object on all instances due to a bug, this has been fixed.
• The rate and voice parameters in the speak command were not optional and this has now been fixed (as has the manual heading)

There are two breaking changes which include the new syntax of the match statement has now changed and the copyof keyword has been dropped. Both of these lead to YASS 24.7. Make sure you read the changelog to understand what it is that has changed.

Namespaces are finally here with ZPE 1.12.8! They are essential for organising libraries and structures throughout programs. Look at the code below to see how they work:

structure HTMLBuilder

  namespace jsmith/packages

  /*
  * Functions go here
  * ...
  */

end structure

$builder = new jsmith/packages/HTMLBuilder()
  

I usually come up with codenames for the next version of ZPE quite far in advance, and this year is no exception. 

For version 1.13, the following codenames have been decided:

1. Petergate
2. Micklegate
3. Davygate
4. Castlegate
5. Monkgate
6. Goodramgate
7. Fossgate
8. Colliergate
9. Gillygate
10. Walmgate
11. Coppergate
12. Stonegate

The inspiration for these codenames came from the City of York, where I recently had a short break.

ZPE 1.12.7 is now available to download. This version brings sweeping changes to the way that libraries are imported into YASS code and the ZRE; it introduces bubbling, brings breakpoints to ZPE, fixes an issue with permission levels not being read correctly, and, perhaps most importantly, adds unescaped string literals to the YASS language.

All of these features makes ZPE 1.12.7 one of the most jam-packed releases of ZPE to date.

I have been seeking to add breakpoints to ZPE for the last seven years, and I have finally succeeded. 

They work using the new bubbling technique I have developed. In ZPE, bubbles are exceptions; a breakpoint is a simple bubble. Bubbles go all the way to the top of the runtime and are handled by the parent. Complex bubbles work entirely differently, and although they are also bubbles, they can be handled by different parts of the program. 

A breakpoint is added to YASS code by writing #breakpoint# where the code should break. In the GUI mode, it will display a variable watch window:

Another prominent feature brought to ZPE this year is bubbling. ZPE will introduce bubbling with version 1.12.7. It is used for interrupts such as exit requests, errors, and breakpoints. These are called bubbles since they are up to the program's top. So, for example, an exit request no longer exits ZPE; it handles a bubble instead to decide what to do with it (and, of course, that might be to exit, or it might be to close the active thread).

Bubbling is a huge feature that changes many of the underlying implementations of ZPE.

One of the significant updates coming to ZPE 1.12.7 is a major change to how ZPE handles library imports. This post is designed to outline this change so I can reference it from the changelog and GitHub commit. 

Before this version, ZPE would read in library imports from a yex file in the libraries folder. This hasn't changed. The library files themselves need not be recompiled. However, the way they are imported has changed. In the older version of ZPE, a library was imported as a variable with all the structures and classes being defined within them:

import stdLib
print($stdLib->bubble_sort([2, 7, 1, 6]))

In ZPE 1.12.7, this changes. Now, instead of adding the bubble_sort function to the $stdLib object, the bubble_sort is put into the top level of the runtime (i.e. added to the global function), so instead you would call bubble_sort:

import stdLib
print(bubble_sort([2, 7, 1, 6]))

ZPE 1.12.7 also allows doing this in the old manner by using the new import as syntax:

import stdLib as stdLib
lib = new stdLib()
print(lib->bubble_sort([2, 7, 1, 6]))

This new change will make some difference to the way in which libraries are used and will improve ZPE performance and memory footprints.