Please review the current release notes of LiveCode 4.6 Desktop versions - for Mac OS X, Windows and Linux.
LiveCode 4.6 is a new release featuring a range of improvements and new features. This document describes all the changes that have been made – including bug fixes and new syntax.For information about improvements made to the iOS port of the engine, please see the iOS Release Notes PDF.
At the time of writing, this information has yet to been integrated into the dictionary or User's Guide.
The engine supports a variety of operating systems and versions. This section describes the platforms that we ensure the engine runs on without issue (although in some cases with reduced functionality).
The engine supports the following Windows OSes:
Note: On 64-bit platforms the engine still runs as a 32-bit application through the WoW layer.
The linux engine requires the following:
Note: The optional requirements (except for gksu and lcms) are also required by Firefox and Chrome, so if your linux distribution runs one of those, it will run the engine.Note: If the optional requirements are not present then the engine will still run but the specified features will be disabled.Note: LiveCode and standalones it builds may work on remote Xservers and in other bit-depths, however this mode of operation is not currently supported.
The Mac engine supports:
Note: The engine runs as a 32-bit application regardless of the capabilities of the underlying processor.
The following changes are likely to occur in the next or subsequent non-maintenance release:• The engine (both IDE and standalone) will require gtk, gdk, glib, pango and xft on Linux
The structure of the IDE install has changed significantly in this release. Each distinct version has its own complete folder – multiple versions will no longer install side-byside: on Windows (and Linux), each distinct version will gain its own start menu (application menu)entry; on Mac, each distinct version will have its own app bundle.
The default location for the install on the different platforms when installing for 'all users' are:• Windows: /RunRev/ LiveCode 4.6• Linux: /opt/runrev/livecode-4.6• Mac: /Applications/ LiveCode 4.6.app
The default location for the install on the different platforms when installing for 'this user' are:• Windows: /RunRev/Components/LiveCode 4.6• Linux: ~/.runrev/components/livecode-4.6• Mac: ~/Applications/ LiveCode 4.6.app
Note: If your linux distribution does not have the necessary support for authentication (gksu) then the installer will run without admin privileges so you will have to manually run it from an admin account to install into a privileged location.
On Windows, the installer hooks into the standard Windows uninstall mechanism. This is accessible from the appropriate pane in the control panel.On Mac, simply drag the app bundle to the Trash.On Linux, the situation is currently less than ideal:• open a terminal• cd to the folder containing your rev install. e.g.
• execute the .setup.x86 file. i.e.
• follow the on-screen instructions.
If you find that the installer fails to work for you then please file a bug report in the RQCC or email
In the case of failed install it is vitally important that you include the following information:
• Your platform and operating system version• The location of your home/user folder• The type of user account you are using (guest, restricted, admin etc.)• The installer log file located as follows:• Windows 2000/XP: //Local Settings/• Windows Vista/7: //AppData/Local/RunRev/Logs• Linux: /.runrev/logs• Mac: /Library/Application Support/Logs/RunRev
The license system has been replaced in this release.
The new system ties your product licenses to a customer account system, meaning that you no longer have to worry about finding a license key after installing a new copy of LiveCode. Instead,you simply have to enter your email address and password that has been registered with our customer account system and your license key will be retrieved automatically.
Alternatively it is possible to activate the product via the use of a specially encrypted license file. These will be available for download from the customer center after logging into your account. Thismethod will allow the product to be installed on machines that do not have access to the internet.
In order to better support institutions needing to both deploy the IDE to many machines and to license them for all users on a given machine, a number of facilities have been added which areaccessible by using the command-line.
Note: These features are intended for use by IT administrators for the purposes of deploying LiveCode in multi-user situations. They are not supported for general use.
It is now possible to invoke the installer from the command-line on both Mac and Windows. When invoked in this fashion, no GUI will be displayed, configuration being supplied by argumentspassed to the installer.
On both platforms, the command is of the following form:
install noui options
Here options is optional and consists of one or more of the following:
Note that the command-line variant of the installer does not do any authentication. Thus, if you wish to install to an admin-only location you will need to be running as administrator before executing the command.
As the installer is actually a GUI application, it needs to be run slightly differently from othercommand-line programs.In what follows should be replaced with the path of the installer executable or app(inside the DMG) that has been downloaded.
On Windows, you need to do:
start /wait install noui options
On Mac, you need to do:
“/Contents/MacOS/installer” install noui options
On both platforms, the result of the installation will be written to the console.
In a similar vein to installation, it is now possible to activate an installation of LiveCode for allusers of that machine by using the command-line. When invoked in this fashion, no GUI will be displayed, activation being controlled by any arguments passed.
On both platforms, the command is of the form:
activate -file license -passphrase phrase
This command will load the manual activation file from license, decrypt it using the given passphrase and then install a license file for all users of the computer. Manual activation files can be downloaded from the 'My Products' section of the RunRev customer accounts area.
This action can be undone using the following command:
Again, as the LiveCode executable is actually a GUI application it needs to be run slightly differently from other command-line programs.In what follows should be replaced with the path to the installed LiveCode executable or app that has been previously installed.
start /wait activate -file license -passphrase phrase
start /wait deactivate
“/Contents/MacOS/LiveCode” activate -file license -passphrase phrase
On both platforms, the result of the activation will be written to the console.
The import and export snapshot commands have been augmented with the ability to specify whether graphic effects and the blendLevel should be applied to the object before rending.
The revised syntax of these commands is:
import snapshot from [ rectangle rectangle of ] object [ ( with | without ) effects ]
export snapshot from [ rectangle rectangle of ] object [ ( with | without ) effects ] …
Where … denotes the usual possible suffixes for the export snapshot command.
If the effects clause is not specified, then the object is rendered without effects.
When rendering with effects rectangle is clipped to the maximum rectangle the object needs to fully render all the applied effects (this may be considerably bigger than the object, is a distant dropshadow is involved).
The syntax for the 'print link' command has been changed to:
print link to [ url | anchor ] link with rect rectangle
If the url adjective is specified, then link will be interpreted as a URI and will create an external URI launching link in the PDF document.
If the anchor adjective is specified, then link will be interpreted as an internal anchor name and will create an internal anchor link in the PDF document.
If no adjective is specified then link will be interpreted as a URI if it begins with (ASCII) letters followed by a colon. e.g. “mailto:foobar” would be interpreted as a URI, whereas “mail-toanchor:foo” would be interpreted as an anchor name.
You can now create a document outline in PDFs generated by the pdf printing commands.Individual entries in the document outline are referred to as 'bookmarks' and can be created with the following syntax:
print [ unicode ] bookmark title [ with level level ] [ at location ]
Here title is the name of the bookmark which will be displayed, level is the depth at which the entry should be placed, and location is a point describing where on the current page the bookmark should refer to.
If the unicode adjective is specified, then title will be interpreted as a UTF-16 encoded string, otherwise it is interpreted as the native (single-byte) text encoding.
If level is not present, then the entry is inserted at the level of the previously inserted item, defaulting to the top-most level if no previous entries have been made.If location is not present, the top-left of the current page will be taken as the location.
Note: The final syntax for this command is still being considered as there are other options we'd like to include, therefore it might change in a subsequent release but will be final for release of 4.6.
Several field properties related to plain text export have been added or updated in the field object.
The formattedText and unicodeFormattedText properties return the contents of the field as plain text after any wrapping has been performed, where soft (or explicit) line-breaks are exported as a singlereturn character, and paragraph-breaks are exported as two return characters. Any paragraphs with listStyle set get prefixed by an appropriate plain-text form of the bullet or index.
The plainText and unicodePlainText properties return the contents of the field as plain text with any listStyle properties being converted appropriately into plain-text for the paragraphs they affect.
In both cases the unicode property variants return text encoded in UTF-16 in host byte-order, where as the non-unicode properties return the text encoded in the native platform encoding (with '?'replacing any non-convertible characters).
It is now possible to configure a group to turn off automatic clamping of scroll offsets when scrollbars are not present. To control this use the following properties:
set the unboundedHScroll of group to booleanValue
set the unboundedVScroll of group to booleanValue
When an unbounded scroll property is set to true and the corresponding scrollbar is not visible, it is possible to set the scroll for that axis to values outside that of the size of the content (both positive and negative).The main motivation for this change is to allow much easier implementation of the 'bouncing' effect common in iOS controls. Indeed, after setting the relevant unbounded scroll properties, it is enough to just use the scroll value provided by a scroller object directly as the value passed to the relevant group scroll property.
If a scrollbar is made visible or unbounded scroll is turned off and the corresponding scroll property is outside of the content bounds, the group is scrolled to bring the value within the appropriate range.
A randomBytes() function has been added to enable generation of arbitrary amounts of cryptographic-quality random data. To use this feature, the syntax is:
The function will return byteCount random bytes and uses the OpenSSL library's random data generator to do so.
The random data is derived from non-predictable sources where possible, meaning that it can (for the most part) be considered truly random. This is contrast to the random function which uses a pseudo-random number generator and randomSeed value.Note: Make sure the security library is included when building applications that use this function as a standalone. If the library cannot be found at runtime, the function with throw an error.
The SHA-1 digest of a block of data can now be computed by using the following form:
the sha1Digest of data
This function returns the sha-1 digest in the form of 20 (binary) bytes.
The encrypt and decrypt commands now support RSA public key encryption. The new forms supporting this are:
encrypt message using rsa with ( public | private ) key key [ and passphrase passphrase ]
decrypt message using rsa with ( public | private ) key key [ and passphrase passphrase ]
Here key should be in PEM format, optionally protected by passphrase.The maximum length of a message that can be encrypted using RSA is the size of the key in bytes – 11. So, for a 512-bit key pair, the maximum encryptable message size is 53 bytes.
Public-private key pairs can be generated using the OpenSSL suite of command-line tools. Forexample:
openssl genrsa -out private_key.pem 512
openssl rsa -pubout -in private_key.pem -out public_key.pem
Will generate a key pair of size 512-bits, placing the private key in private_key.pem and the public key in public_key.pem.For more information on these utilities see http://www.openssl.org/docs/apps/rsa.html and http://www.openssl.org/docs/apps/genrsa.html.
Use encrypt with a public key to encode a message that you only want to be decoded the holder of the private key.
Use decrypt with a private key to decode a message that a sender has encrypted with its corresponding public key.
Use encrypt with a private key to encode a message that a receiver can then verify has come from one of the holders of the private key (this is a signing operation).
Use decrypt with a public key to verify that a message has been encoded with the corresponding private key, and there has come from one of its holders (this is a verify operation).
For signing, the maximum length of an encryptable message isn't really an issue since typically in that scenario it will be some sort of hash that would be being encrypted.For the more traditional encrypting scenario, however, the standard approach is to use public key cryptography to encrypt a random password which is then used with a symmetric cipher to actuallyencrypt the payload.
Standard Mac document windows typically display the modified state of the document they represent by adding a dark dot to the close-box. It is possible to control the display of this dot in LiveCode by using:
set the modifiedMark of stack to booleanValue
The indicator will be displayed when the property is set to true.This property has no effect on Windows nor Linux.
Support has been added to some revXML functions for manipulating text nodes. Consider the following XML fragment:Removes a message that was queued with thesend command and is waiting to be sent.
Removes a message that was queued with thesend command and is waiting to be sent.
This has the following structure as an XML tree:
TEXT( that was queued with the )
TEXT( command and is waiting to be sent.)
Notice that the named XML nodes, are interspersed with (essentially unnamed) nodes containing text content. These (previously unaccessible) nodes can now be accessed (with some functions) by using a path of the form:
Here this references the th text node under . For example, the above tree has the following accessible nodes:
To access text content of a node simply use the revXMLNodeContents function with this extended path format. Note that the previous behavior is preserved – if you specify a node with a '/[n]' suffix,the text content of the node will be returned (if it is a text node). (i.e summary/keyword and summary/keyword/ are the same thing to revXMLNodeContents).
The following functions have been augmented with an additional (optional) parameter incText:
revXMLFirstChild(docId, [ incText ])
revXMLNextSibling(docId, nodePath, [ incText ])
revXMLPrevSibling(docId, nodePath, [ incText ])
revXMLChildNames(docId, nodePath, delimiter, filter, incCounts, [ incText ])
Here if incText is specified and is true, the functions will include text nodes in their processing.
For example, this allows you to loop over all nodes including text nodes using something like:
put revXMLFirstChild(tDocId, tParentNode, true) into tCurrentNode
repeat while tCurrentNode is not empty
... use tCurrentNode ...
put revXMLNextSibling(tDocId, tCurrentNode, true) into tCurrentNode
Important: This feature is currently experimental. This means that it may not be complete, or may fail in some circumstances that you would expect it to work. Please do not be afraid to try it out as we need feedback to develop it further.
Experimental support has been added to the field for simple, single level, ordered and unordered lists. To make a paragraph display as an element in a list use the listStyle property:
set the listStyle of line lineIndex of field to style
Where style is one of:
• disc – the paragraph is rendered as an element in an unordered list with the standard bullet character as marker (U+2022).• circle – the paragraph is rendered as an element in an unordered list using the (unicode) character U+25E6 as marker.• square – the paragraph is rendered as an element in an unordered list using the (unicode) character U+25AA as marker.• decimal – the paragraph is rendered as an element in an ordered list, the label using standard (Arabic) decimal numerals. i.e. 1, 2, 3, 4, etc.• lower latin – the paragraph is rendered as an element in an ordered list, the label using lowercase (Latin) letters. i.e. a, b, c, …, aa, ab, etc.• upper latin – the paragraph is rendered as an element in an ordered list, the label using uppercase (Latin) letters. i.e. A, B, C, …, AA, AB, etc.• lower roman – the paragraph is rendered as an element in an ordered list, the label using lowercase Roman numerals. i.e. i, ii, iii, iv, …, etc.• upper roman – the paragraph is rendered as an element in an ordered list, the label using lowercase Roman numerals. i.e. I, II, III, IV, …, etc.
For both ordered and unordered list the marker is placed at the first tab-stop, and the content of the paragraph is placed (and wraps) at the second tab-stop.
For ordered lists the index of the item is determined by the number of preceding paragraphs withthe same listStyle property.
Setting the listStyle property of a paragraph to empty causes it to revert to a normal (non-list item) paragraph.
Paragraphs with a non-empty listStyle present themselves in htmlText wrapped with
tags.Sequences of such paragraphs with the same listStyle are bracketed by 2. Numbered Item 2
Numbered Item 1
• Bulleted Item 1
Numbered Item 2
Bulleted Item 1
Bulleted Item 2
tags. Thesetags take a type attribute, matching the (legacy) HTML attribute of the same name:• disc → disc• circle → circle• square → square• 1 → decimal• a → lower latin• A → upper latin• i → lower roman• I → upper roman
1. Numbered Item 1
• Bulleted Item 2
Paragraphs with a non-empty listStyle are appropriately marked in rtfText using both the (legacy) pn family of paragraph numbering tags and also with the new listtable tags.
By using both sets of tags a reasonable degree of interoperability is achieved with both TextEdit (and other Cocoa applications) on Mac, and Word and WordPad on Windows.
Note: Unfortunately, OpenOffice does not have particularly good rtf import / export capabilities (it doesn't even round-trip correctly through itself!) and thus copying / pasting of lists betweenLiveCode and OpenOffice will not work reliably or correctly.
Paragraphs with a non-empty listStyle are appropriately exported in plain text form when using the plainText, unicodePlainText, formattedText and unicodeFormattedText properties.
For example, the above example would be rendered (for plainText) as:
1.Numbered Item 1
2.Numbered Item 2
•Bulleted Item 1
•Bulleted Item 2
As it stands the listStyle property does not save into the stack-file and will not do so until version 5.0 when the file format is revised (for various technical reasons, it is not possible to add this as a saveable property with the current stackfile format).
It is recommended that htmlText be used to save the content of fields in custom properties, for restoration on reload.
Important: This feature is currently highly experimental. This means that it may not be complete, fail in some circumstances that you would expect it to work, or change considerablybefore becoming final. Additionally, it might have problems or gotchas that make it significantly harder to use than other LiveCode features at this time.
By default, revXML will process XML Namespace related tags and attributes in a way that is inaccessible to revXML. This can cause problems when a script needs to process the namespaces itself.
To resolve this issue two additional functions mimicking the behavior of revCreateXMLTree and revCreateXMLTreeFromFile have been added to revXML. These functions are:
The difference between these and the originals is that these two functions ignore namespace tags and attributes, and instead return them as part of the tree without doing any processing.
For example, consider the following XML tree:
Will get interpreted as follows
The latter form is more appropriate in the case that an application needs the namespace information to perform correctly.
In order to be able to more reliably control the maximum level of recursion, a new global property stackLimit has been introduced.
This property allows a script to set (in bytes) the maximum size of the (runtime) stack the engine uses for recursive computation. A change in the setting will only take effect when all currently executing handlers complete, and at this time the stack size limit will be reconfigured to the given limit, or the nearest amount to it depending on available memory.
The stackLimit currently in effect can be fetched using the effective stackLimit.
The recursionLimit property is now bounded by the stackLimit – attempts to set the recursionLimit greater than the stackLimit will see it downwardly adjusted to the maximum current size allowed.
Note: The changes to the recursionLimit property and the new stackLimit property are only implemented on Windows at present.
There is a long standing issue with revBrowser that causes browser instances to be lost whenever the stack it is attached to has its window re-created. Previously, cases where this would occur had to be avoided when a browser was present on a stack.
To resolve this problem a new property has been added to browser instances – windowId. The windowId property allows the stack to which a browser instance is attached to be changed after ithas been created.
If the windowId is set to 0, the browser instance is temporary hidden. If the windowId is set to a valid stack windowId, the browser instance will move to that stack.
For example, to toggle the resizable property of a stack hosting a browser use the following code:
revBrowserSet pBrowserId, "windowId", 0
set the resizable of stack pBrowserStack to pNewResizeableValue
revBrowserSet pBrowserId, "windowId", the windowId of stack pBrowserStack
Important: This feature is currently experimental. This means that it may not be complete, or may fail in some circumstances that you would expect it to work. Please do not be afraid to tryit out as we need feedback to develop it further.
The open socket command no longer blocks on DNS resolution. Instead, if resolution is required the command will return immediately and the DNS lookup will happen in the background. If resolution fails, then a socketError message is sent in the same was as if connection fails.
For applications using hostNameToAddress directly, its syntax has been augmented:
hostnameToAddress(hostname, [ callback ])
If the callback parameter is specified then the call will return immediately and upon completion of the lookup, the callback will be invoked with the resolved address as a parameter.
Cursor support has been improved in several ways.
The image will now automatically process any image to convert to a form suitable for display on the current platform and screen depth taking this burden off the developer. Specifically, the enginewill scale the image down to the appropriate size, and reduce the number of colors to the appropriate number of colors.
Additionally, the engine has been updated to take advantage of support for larger and potentially alpha-blended cursors on platforms that support this. Specifically:• Windows XP and above support full alpha-blended cursors up to 64x64• More recent Linux distributions support alpha-blended cursors up to a size of 64x64• Mac supports alpha-blended cursors up to 256x256
The engine now picks up the GTK cursor theme on Linux. (4.5.1)Note: It appears that alpha-blending support depends on the current screen depth on some platforms.
The export command can now produce Windows BMP format images in the same was at it previously did for gif and png:
export target as bmp
The export command can now perform color reduction. To support this, the following new forms have been added:
export target as ( gif | png | bmp ) with palette colors
export target as ( gif | png | bmp ) with ( standard | optimized ) palette
export target as ( gif | png | bmp ) with count color optimized palette
The first form allows you to specify a list of up to 256 colors to use in the final palette.
The second form will use either the standard 'websafe' palette, or will compute an optimized palette with at most 256 colors.
The third form allows you to choose the size of the optimized palette to be generated. The number of colors can be at most 256.
If the image has any sort of transparency, then this will utilize one palette entry. (In particular if you ask for a 256 color palette with a transparent image, you may get only 255 colors).
In all cases, optional dithering will be performed as determined by its dontDither property.
Note: It is a (relatively) slow process to compute an optimized palette and then remap an image against it – it should not be considered a real-time operation except for very small images.
Note: There is no support for exporting a JPEG with a reduce palette as this format is for continuous-tone images and as such the notion of palette makes no sense.
It is possible to export raw image data using the following forms:
export target as raw with palette colors
export target as raw with ( standard | optimized ) palette
export target as raw with count color optimized palette
export target as raw [ argb | bgra | abgr | rgab ]
The first three of these operate in the same way as for the other formats as described above except that instead of formatted image data you get the raw palette indices packed appropriately dependingon the size of the palette:
<= 2 colors will be 1 bpp
<= 4 colors will be 2 bpp
<= 16 colors will be 4 bpp
<= 256 colors will be 8 bpp
The final form allows export of the full 32-bit data of the image with 8 bits per component. In this case, the components are not pre-multiplied with any alpha channel, and appear ordered in memoryin increasing bytes.e.g. The argb form will give you:byte 0 = alphabyte 1 = redbyte 2 = greenbyte 3 = blue
Important: This feature (raw data export) is currently experimental. This means that it may not be complete, or may fail in some circumstances that you would expect it to work. Please donot be afraid to try it out as we need feedback to develop it further.
The engine will now attempt to use any embedded ICC color profile information that is present in JPEGs and PNGs.
If a JPEG contains a color profile then the engine will attempt to use it to translate the image's colors to the default screen color space before display.
When printing JPEGs, the original JPEG data will now be sent to the printer directly whenever possible, In particular, this means that any intermediate color matching for the screen that the engine performs does not affect the printed output.
As an additional side-effect of color profile support, the engine now also supports YCCK and CMYK JPEG images.
Note: Support for color profiles depends on the OS - in particular Windows XP and earlier only support ICC v2 profiles, Vista onwards supports ICC v4.
The engine printing system has been augmented with the ability to print to pdf. This feature uses the existing print-loop model and virtually no code changes are needed to make existing code use it.
To start a print loop that outputs directly to a pdf rather than the currently configured printer, use the following form:
open printing to pdf filename [ with options optionArray ]
This should be performed instead of the usual open printing command.
Note: It is not necessary (and unwise!) to set the formatForPrinting option on stacks which arebeing printed to PDF.
The options array which can be optionally specified when opening a print loop for pdf allows you to add entries to the resulting PDF's Document Information Dictionary.The following keys are supported:
Title, Author, Subject, Keywords, Creator, Producer
Their values can be any string.
When inside a pdf print loop you can use additional commands to define hyperlinks and their targets.
To define a target for an internal hyperlink, use the following form:
print anchor name at anchorPoint
Here name is used to identify the anchor in related print link commands, and anchorPoint is the location on the current page to which any such link should jump to.To define a hyperlink use the following form:
print link to target with rect linkRectangle
If target is a url, then the given rectangle will jump to that url when it is clicked. If target is not a url, it is assumed to be the name of an internal anchor as specified by the print anchor command and clicking in the given rectangle will jump there instead.
When printing fields, any text that has its linkText property set and has the textStyle link set will be treated as if a print link command had been executed with the contents of the property as target,and the formattedRect of the text as rectangle.
The implementation of pdf printing resides in a dynamic library called revpdfprinter. The standalone builder will automatically include this in the appropriate place when building standalones.
Any custom code that deals with standalones which use this feature must make sure that the revpdfprinter library resides next to the engine executable (On Mac, this is inside the Contents/MacOS folder, not next to the bundle).
A number of issues with the open process command and the engine itself have, up until now, conspired to make it difficult (if not impossible!) to either run a slave process, or use the engine as slave on all platforms.
These issues have been resolved in this version, thus making it straightforward to run another process and poll for input and output over stdin/stdout.
The typical form for this is along the following lines (this example assumes the process being executed outputs whole lines):
command startSlave pProcess
open process pProcess for text update
send “monitorSlave pProcess” to me in 50 millisecs
command monitorSlave pProcess
# Loop until there are no more lines to read.
read from process pProcess for 1 line in 0 millisecs
if the result is empty then
# The slave has sent us something, so process it and loop for
# (potentially) more data.
else if the result is “timeout” then
# There is nothing waiting for us, so exit repeat
else if the result is “eof” then
# The slave has terminated, so do any final processing and finish
close process pProcess
# Some error has occurred!
send “monitorSlave pProcess” to me in 50 millisecs
Some work has been done on storing and fetching settings from the videograbber.
On Windows and Mac, audio and video settings are now correctly retrieved and set using revVideoGrabSettings and revSetVideoGrabSettings.
On Windows (when using DirectShow), the 'Camara Control' settings are now saved and restored via the grab settings commands. Specifically: pan, tilt, roll, zoom, exposure, iris, focus and flash.
Sometimes it is necessary to perform operations on the local machine as an administrator, and a typical pattern for a GUI application doing this is for it to prompt for authentication at certain points.
Modern operating systems do not permit a process to elevate itself, nor grant itself increased privilege. Instead, they only allow a running process to launch another process with increased privilege. Therefore, in order to support this, a new form of the open process command has been introduced that can launch a slave process with elevated permissions:
open elevated process process [ for [ text | binary ] ( read | write | update | neither ) ]
This form operates identically to the normal version, except that engine will ask the system to launch the given process with admin/root privileges.
The standard way for a GUI application that needs to perform privileged operations to be structured is to split the application into two parts: a GUI front-end that interacts with the user, and a command-line back-end that is run with elevated permissions. These two parts can then talk to each other using a standard master-slave approach, or some other form of IPC such as sockets.
To get a list of sub-keys in the Windows registry use the following function:
This will return a return-delimited list of sub-keys, i.e. those keys which are direct children of the given parentKey. The specified key should be in the same format as the other registry functions.
In previous versions it was necessary to set the sslCertificates property to the root certificates that HTTPS connections should be verified against. Support has now been added to locate and load theroot certificates installed (and kept up to date) as part of the OS.
This uses the standard root certificate keychain on Mac, the standard root certificate store on Windows and uses a number of heuristics to locate this information on Linux.You can easily find out if the system-installed root certificates are being found by running the following command in the message box:
get url "https://www.google.com"
put the result & return & it
If this results in an error about verification failure then it is likely that root certificates have not been found. Please let us know (particularly on Linux) if you find this simple test fails, making sure yougive us full details of your system (e.g. Linux distribution and version).
Note: Unfortunately this feature does not currently work correctly on Mac 10.6.x. For now, we advise including an appropriate root certificates collection with your application, as was previouslynecessary, and setting the sslCertificates property appropriately.
Previously unsupported syntax for manipulating the dock icon on Mac is now experimental.
The current dock icon image can be set by using the global icon property:
set the icon to imageId
The engine will attempt to find an image with the given id, resize it to 128x128 and then set it as thedock icon for the application.This property has no effect on other platforms.
Note: The image is only guaranteed to persist while the application runs, although in some cases the OS does appear to cache it beyond this.
In addition to changing the dock icon image, you can also configure the menu that appears when the user clicks on it.
To set the dock icon menu use the global iconMenu property:
set the iconMenu to iconMenuSpec
Here, iconMenuSpec is a string describing the menu. This uses a subset of the standard menu specification syntax. The string should be a return-delimited list of items specified as follows:
[ * ] [ '(' ] [ '|' ]
Here the number of tabs determines the depth of the menu (i.e. use this to create sub-menus). The optional tag is used when calling the iconMenuPick message.Before the engine displays the icon menu, it will send a iconMenuOpening menu to the current card of the defaultStack. You can use this opportunity to change the icon menu before it isdisplayed, this is an analog to handling mouseDown in a menu button.When the user selects an item from the dock menu, the engine will send an iconMenuPick message to the current card of the default stack:
Here which will be a list of labels or tags (if specified) separated by '|' which determines which item was selected.
The following bug fixes or changes are ones which may have an impact on existing code and/or may be important to a wide audience.(changes specific to the current build are highlighted with a gray background)
Mac OS 10.3.9 is no longer supportedSupport has been dropped for Mac OS 10.3.9.
Clipboard format for styled-text on Mac now RTFThe styled text format that LiveCode exports to the clipboard on Mac has been updated to be RTF – this brings it inline with both Windows and Linux, and will allow support for hyperlinks andlistStyles to be supported in a subsequent build.
SQLite updated to 3.7.2 with Full Text Search enabled
The verison of SQLite compiled into dbSqlite has been updated to version 3.7.4. In addition, the 'FTS' feature has been compiled in. For more information about this version of SQLite and the FTSfeature please see the documentation at http://www.sqlite.org.
Non-character keys can cause erroneous keyDown messages on MacPreviously keys such as page-up, page-down and the function keys would cause keyDown messages to be sent to a field but with a nonsensical parameter. These keys no longer generatekeyDown in line with other platforms.
The listStyle property has better RTF import / export supportSupport for both types of list specification in RTF has been added, increasing interoperability with other applications when using the clipboard. See the section on the listStyle feature for moreinformation.
Bug 818 – About item always present and enabled on MacThe engine will now only add an About item to the Apple menu if it finds a menu with name Help whose last item has tag About or label beginning with About.
Furthermore, the about item is enabled only if the found item is enabled and the help menu is enabled.
Bug 3071 – Preferences item cannot be disabled on MacThe engine will now only enable the Apple menu preferences item if the following are all true:• There is a menu with name Edit• The edit menu is enabled• The last item of the edit menu has tag Preferences, or has label beginning with Preferences• The preferences item is enabled
Bug 1197 – backgroundBehavior policy now enforcedThe engine now strictly enforces the following two rules:• a nested group cannot have the backgroundBehavior set to true• a group placed onto multiple cards cannot have its backgroundBehavior set to falseOn loading stacks in the IDE, a consistency check will be performed which will result in these rules holding after load.
Bug 2122 – Styled text overruns tab-stopsYou can now use styled text correctly when a field has the vGrid property turned on. The text will be clipped correctly at cell boundaries.
Bug 4026 – importing images > 4091 wide causes corruption on MacThe long standing limit on the width of image objects on Mac has been lifted. Images are now limited by available memory, rather than the width.
Bug 7334 – rtfText output is now effective
The rtfText property is now 'effective' in the sense that it contains styling for all characters, including those which inherit text properties from the field (or more distant ancestor). This change has been made to make it more useful for clipboard export. If you wish to fetch the contents of the field with the inheritance structure intact, use the htmlText instead.
Bug 9301 – menuHistory now has consistent behavior for all menu button types
Previously, setting the menuHistory would only work reliably for option/combo buttons. This has been rectified and setting the menuHistory (or using select menuitem … of menu …) will now cause the correct menuPick message to be dispatched for all menu button types.
Note that for option and combo buttons, such a message will only be sent if the menuHistory changes, this is consistent with these objects having a notion of 'currently selected item'.
Bug 9378 – 'mailto:' URLs not recognized by 'print link to'In order to better support usage, the syntax and functionality of the 'print link' command has changed. In particular, the print link command now accepts a 'url' adjective to better specify the typeof link. This will require changes to existing scripts in a very specific case.
Code such as:
print link to url “file:mylinktext.txt” with rect tMyLinkRect
Needs to be changed to:
print link to (url “file:mylinktext.txt”) with rect tMyLinkRect
Bug 9412 – Messages passing through a background are received twice if passedThe message path order for backgrounds has been revised slightly (in a backwards-compatible fashion) to take proper account of passing of messages. For a control within a background, themessage path is the following:
target's owning groups
card's backgrounds not including the target's background
The path for a message sent to a background is:
Note: The above order effectively means that a background's script is only behind the card for controls it does not contain which retains compatibility with previous interpretations of backgroundBehavior.
(bug fixes specific to the current build are highlighted in bold, reverted bug fixes are strickenthrough)
Text linkClicked parameter contains an incorrect trailing character in some cases9466 Tab key does not insert a tab into fields.
Non-character keys can cause erroneous keyDown messages on MacFonts in styled text can paste with incorrect style on Mac in some cases'ask folder' uses new style dialog on Vista and later'go as sheet' doesn't disable the parent window on Windows
tags are not optional when in
9368 CMYK JPEGs don't appear in printed PDFs9375 The 'templateField' does not convert text to htmlText9377 mouseDoubleUp not sent to locked and list fields9378 'mailto:' URLs not recognized by 'print link to'9380 listStyle doesn't report 'mixed' when it should in many cases9381 'export snapshot' does not render anti-aliased text correctly on Windows9382 Radio buttons with custom icons don't update correctly when in backgounds9383 Crash when setting 'the backgroundBehavior' of the templateGroup9384 Unnecessary limit on the number of menu buttons on open cards on Mac9388 Change in how htmlText is interpreted9395 Crash when setting field rect on Linux9396 Tooltips don't have drop-shadows on Windows 79399 Can't get the short name of the dragSource9400 'the text of the selectedLine' is always empty9403 Right-click then Cut on Mac causes crash on empty image9404 The formattedHeight of a field counts the scrollbarWidth twice9410 Current external load process is dependent on current folder on Windows9412 Messages passing through a background are received twice if passed9413 The ctrlKey synonym is tied to commandKey9417 Only the first link on a line ever activates when listBehavior is true9419 The pixel properties only work if the object is on the current card of an open stack9428 Modal windows do not behave correctly on Linux9435 BMP images with depth < 32 and odd width do not import correctly9439 Text does not render correctly on non-primary screen on Linux (not tested)9447 The linkClicked message does not pass the full linkText value as a parameter9451 Numeric keypad input does not work on Mac
If you write plugins, or have code that relies on the location of IDE files then please ensure you use the following access functions to locate them:revEnvironmentToolsPath() The location containing the main IDE files.revEnvironmentToolsetPath() The location of the main IDE stacks.revEnvironmentExternalsPath() The location of the externals that come with the IDE.revEnvironmentPluginsPath() The location of the plugins that come with the IDE.revEnvironmentRuntimePath() The location of the standalones that come with the IDE.revEnvironmentDocumentationPath() The location of the documentation files.revEnvironmentResourcesPath() The location of the resources that come with the IDE.revEnvironmentCustomizationPath() The location of the IDE customization folder.revEnvironmentUserCachePath() The location of the folder to use for caching files.revEnvironmentUserPreferencesPath() The location of the folder to use for preference files.revEnvironmentUserExternalsPath() The location of the folder to use for additional externals.revEnvironmentUserPluginsPath() The location of the folder to use for additional plugins.revEnvironmentUserResourcesPath() The location of the folder to use for additional resources.
Important: Third-party IDE extensions must avoid placing any files inside the application bundle or under revEnvironmentToolsPath() (not least because you will probably not have privileges to do so!). Instead, they should use the user-externals and user-plugins paths as provided. These paths are determined by the user's customization path setting, configurable in the preferences.
The update checker and its corresponding menu item in Help have been removed. The version checking method is changing, and this option will re-appear in a subsequent release.
The affiliate registration item in the Help menu has been removed The method of affiliate registration is changing, and this option will re-appear in a subsequent release.
Use of the database query builder has been deprecated.
To access its functionality, choose the Enable database query builder option on the Compatibility & Updates pane of the preferences.
Note: The database query builder will be removed entirely in the next or subsequent nonmaintenance release.
The IDE now contains built-in support for simulating and deploying iOS applications. The following features have been added to support this:• An iOS pane has been added to the Standalone Builder• Save as Standalone Application can now build iOS applications• A Simulate button has been added to the menubar to enable fast and easy access to running applications in the simulator• Configuration of simulator version and type can be accessed via a Simulator Version submenu in the Development menu.• A Mobile Support pane has been added to the preferences to enable configuration of iOS
More details on how to use these features can be found in the iOS Release Notes accessible via the Help menu.Note: If you do not have an iOS deployment pack, you can still use the iOS feature in trial mode. In this case the applications you build will contain a forced banner on startup lasting 5 seconds, andthey will quit after one minute.
The IDE comes with a Resources folder containing the sample projects, example stacks and other miscellany.Previously this was easily accessible on all platforms by navigation to the installed IDE folder. As this folder is now (on some platforms) inside an application bundle, a new item in the Help menuExample Stacks and Resources has been added to open up the folder in the file manager on the running system.
It is now possible to specify what action UAC should take on Windows Vista and higher when the standalone is launched. You can choose one of the following options:
Using the 4.5 standalone builder to build for Web will create revlets. These are compatible with the existing revWeb plug-in available from revweb.runrev.com.
An updated version of the IDE for producing LiveCode Applets and the associated LiveCode player will be made available in due course.
Note: As the currently available revWeb plug-in uses the 4.0 engine, you must be careful to only use features that are present in that version.
There is now a pane for configuring iOS applications in the Standalone Builder. At this time, iOS standalone building has the following limitations (compared to desktop/web deployment);• A stack configured for iOS cannot also deploy to other platforms• The Copy Referenced Files feature is not implemented for iOS builds.• The Inclusions and Property Profiles feature is not implemented for iOS builds.• The Stacks configuration options are not available for iOS builds.
More details on how to use the iOS settings pane can be found in the 'iOS Release Notes', accessible via the Help menu.
The datagrid is currently at version 1.0.2 (build 13).
Added specific scrollbar width for linux platform. DeleteFieldEditorAndOpenNext nowcontinues looking for a column to edit if EditValue is passed in the behavior script. This allows the developer to skip a column for editing. Previously this would only occur if EditValue was not handled. DeleteFieldEditorAndOpenNext now skips invisible columns. The control that DeleteFieldEditorAndOpenNext was targeting would sometimes be changed behind the scenes. This would cause CloseFieldEditor to be sent to the wrong control.
The following bug fixes are ones which may have an impact on existing code and/or may be important to a wide audience.(changes specific to the current build are highlighted with a gray background)
After observing a number of users using LiveCode for the first time, the following minor changes have been made:• The default state of the paint tools in the tools palette is opened so they are visible on start up (existing installs are unaffected).
• The 'Hide Palettes when editing scripts' and 'Hide Message Box when editing scripts' options are now defaulted to false (existing installs are unaffected).• On the menubar the 'Rev Online' button has been renamed 'User Samples' and a new 'Tutorials' button introduced linking directly to http://lessons.runrev.com.
(bug fixes specific to the current build are highlighted in bold) Added shortcut for 'Paste Unformatted' Mac field shortcuts do not work correctly in the Script Editor The dictionary does not open correctly in some cases. Pressing option in some IDE fields cause them to revert the field. New datagrids are not created properly and can't have their Row Template's shown. LiveTalk still present in the first pane of the preferences GUI. Simulator menu does not filter list by available engines4864 Property inspector's treatment of link style properties does not reflect reality7250 'edit custom prop name' dialog resizable when it shouldn't be8576 IDE processes openDoc filename incorrectly8775 Continuations have an inconsistent indent in the S/E9291 Body text in the about dialog is clipped initially9292 Simulate button does not have a tooltip9299 Substacks are unsorted in the app browser9316 CFBundleVersion not set when building Mac standalones9453 Dictionary note rating broken
A substantial number of dictionary entries have had minor tweaks and fixes applied in this release, most as a result of bug reports in the quality control center. Additionally, new syntax entries have been added for much of the new syntax added since 4.0.
The Mobile Examples are a collection of stacks demonstrating various features specific to the mobile platforms LiveCode supports.
These are accessible from Help > Example Stacks and Resources and are within the Mobile Examples folder.The following changes have been made in this release:
Sound ExampleThe example queue has been updated to use the changed functionality of iphonePlaySoundOnChannel in next mode when there is no sound playing (i.e. to prepare thesound to ensure it plays with zero latency when needed).
To support this change, a Pause and Resume button have been added mapping to iphonePauseSoundOnChannel and iphoneResumeSoundOnChannel.
4.5.0 Release Notes http://www.runrev.com/downloads/livecode/4_5_0/LiveCodeNotes-4_5_0.pdf4.5.1 Release Notes http://www.runrev.com/downloads/livecode/4_5_1/LiveCodeNotes-4_5_1.pdf4.5.2 Release Notes http://www.runrev.com/downloads/livecode/4_5_2/LiveCodeNotes-4_5_2.pdf4.5.3 Release Notes http://www.runrev.com/downloads/livecode/4_5_3/LiveCodeNotes-4_5_3.pdf