>> home >> support >> anthracite 1.4 change log

anthracite support

2/6/2006


Anthracite Version 1.4 Change Log

Anthracite 1.4 adds new functionality, updates existing capabilities, and fixes several user and QA reported issues.


Important Upgrade Notes about Anthracite 1.4:

Anthracite 1.4 requires MacOS X 10.4 (Tiger) or newer.

This version updates the file format, and therefore, documents created with version 1.4 are not usable by older versions of the software.

Certain updates to the functionality of existing processes may require users to revisit and update previously created Anthracite documents.


New Features!

First, the fun part...Anthracite 1.4 adds several new capabilities:

New Sources:

SOAP - Easily access web services results within Anthracite. Set a few parameters and away you go, you can even use AppleScript to set SOAP parameters to automate web services retrievals. See the example documents ("SOAP-*" and AppleScript-SOAP) on the Anthracite distribution disk image.

Load If Changed - a much requested feature, Load If Changed only retrieves content from a specified URL if it has changed since the last attempted load. Combined with scheduling and automation of Anthracite processes, this lets you stay on top of the latest content from websites, RSS feeds and other web accessible sources.

YubNub - YubNub is cool, and if you haven't tried it, click thee on this link and prepare to be amazed: http://www.yubnub.org
with the special permission of the creator of YubNub, it's included here as yet another way to get useful input into your Anthracite processes. See the "YubNub-FeedToText" sample document for a nifty example.

PDF - built in PDF to text handling using the Apple PDF Kit tools. Specify a PDF file and the text available in the document is automatically extracted.

New Processor:

XPath - an incredibly powerful tool for extracting data from tagged documents, XPath has the potential to make old school scraping obsolete. Simply specify the tags you want to extract and voila, out pop the strings. Used with XML SOAP results makes this stuff look downright easy. A few example documents are on the distribution disk image in the Goodies folder, but really they have books about XPath and I encourage you to learn all about it, then use it to make your Anthracite process building even easier and more productive.

Other New Items:

Universal Binary - Anthracite version 1.4 is a Universal Binary for both PowerPC and Intel CoreDuo Macintoshes! This version was built on Intel, as in, if you are running it on a PowerPC, it was generated on an Intel machine. Neat. Performance tests vary, but the Intel CoreDuo does quite well against all the G4 based machines in the Metafy labs (G5 performance tests pending). On smaller files, the performance improvements are smaller, such as 1.6X when parsing the King James Bible against the 1.67Ghz PowerBook, but on a larger file (312MB Anthracite document, 61MB source file, 3500 HTML tables, split into two 61MB output chunks), the performance gain was 4.5X for the 2Ghz Intel CoreDuo iMac over the 1.67Ghz PowerBook G4. I'll take it!

Log Anthracite Messages to File - with this preference, all messages in the Log Window are also written to /tmp/Anthracite.log, helpful for debugging or auditing automated processs.

Configurable Default String Encoding - now Anthracite lets the user control the default string encoding. This should prove useful to users of Unicode and other string encodings (e.g., non-English users) who want to work primarily in an encoding other than ASCII.

Updated AppleScript capabilities - SOAP and YubNub are AppleScriptable. See examples on disk for how to specify SOAP parameter records from AppleScript.

Software Updates:

Find Replace edit sheet would not accept empty replacement for previous setting

Multiple updates to multidimensional array handling, formatting, previewing and exporting (May affect certain processes built on old multidimensional array handling.)

String Encodings - updated UTF16 handling, Universal Binary update for processor endianness, default string encoding. If you are getting empty preview documents in your browser or windows full of Chinese characters when you expected something else, be sure to check your Default String Encoding in your preferences.

MySQL Export array trim fix.

Preview uses system default browser instead of Anthracite default browser (was Safari).

Excerpt Array Row/Column selector defaults to row instead of column for simple list support. (This may require an update to user documents that use the Excerpt Processor.)

Major speed improvement in strip processor with large data

Cleanup of log messages.

Table View UI cleanups

Strip Blank Lines now also looks for tabs.

New Sample Documents (SOAP, XPath, YubNub)

Updated license engine.

Fixed Font Menu Bug in Easter Egg (yes, we QA the easter egg)

Known Software Limitations:

NSURLConnection "bug" - Sadly, we also ship this version with a Known Software Limitation (KSL).

Note: This issue appears to only affect Intel CoreDuo and possibly multiprocessor G5 machines. Single processor G4 machines do not appear to be affected.

It's a serious but intermittent issue that can cause Anthracite to crash, but there is a known workaround, albeit at a limitation, hence the KSL. The issue seems to be isolated to the new Intel CoreDuo platform, although research on the internet has found reports of similar errors on multiprocessor G5 machines (additional research available, see also Apple Bug ID 4350489, although we are not saying that this is an OS bug yet until we've done more research on our own code).

In short, it appears that if you attempt to load a local file ("file://" URL) repeatedly, at some point it may cause a crash in the NSURLConnection system library routine. It appears that larger files trigger the crash more quickly. The workaround is to not run a document repeatedly, instead, load it repeatedly and run it once each time. We regret this limitation of the software, and are working to fix this issue as soon as possible.

Again, this only seems to affect the new Intel CoreDuo machines (and possibly multiprocessor PowerPCs) and only intermittently. Users of single processor machines do not appear to be impacted.

[ Metafy Home ]


last update: 2/6/06

Copyright © 2003-2006, All Rights Reserved. Metafy LLC