Random Thoughts

Necessity is no longer the mother of invention, but instead the trigger for some googling... I've just been writing an XSLT transform to generate DOT code where the text labels are C-style strings (which means that backslashes '\' have to be escaped as '\\' to work correctly). An easy way to do this is with a replace function, but the base XSLT specification does not provide this. What would have taken me a week has already been done by the EXSLT project (c.f. replace). Thank you greatly.

There is a lot of downloadable XSLT code at the EXSLT site, but I cannot find any explicit licensing statements. I'd like to think that it is freely available, and not subject to commercial restriction...

Managed to get some work done on ACK tonight and restored the Windows build (VC6) - which had suffered slow bitrot and been broken. I've tried to keep all the different ACK builds functioning, but prior to the release of Minix 3 they were ignored and are most probably in a (mildly) broken state. All of the new functionality (i.e. new machine backends) added since the MinixACK-1.1.2 release has not yet been added to the non-Minix builds and some work will be required to get this all to build. Having the VC6 build is great and increases confidence in the system. I'll now focus on getting the other Windows builds (Borland C v5.5; and command line MSC - LCC-win32; Pelles; and Watcom are not yet supported) functioning, and then concentrate on integrating C-Cured. It will be interesting to see what errors CCured finds in the ACK sources...

XML revisted

It's been a while since the XML books arrived. I've had a quick read of XML Handbook and Definitive XSLT and XPath. The XML Handbook provides a buzzword compliant overview of the XML technologies and has definitely been written with marketing as a priority over substance. Fortunately the XML Core Tutorials (part 15) does provide a very solid technical overview of the different XML technologies and I've learnt a lot from it. There is quite a lot of material covered in the (almost) 300 pages of the tutorials and it is well written - this all makes the book worthwhile.

I've been referencing Definitive XSLT and XPath extensively in the last couple of days now that I'm writing XSLT 'in anger' (and have to get it to work!), but am finding the work to be of less value than I anticipated. The (few) cross-references in the book are useless as they reference sections in the book rather than individual pages. Since each section can be quite long I'm finding that getting information out of the book is rather time-consuming. The book would really profit from having good summaries of the material, but it appears that no effort has been made in providing this. In constrast, the material at w3schools is excellent and I've started referencing their XSLT element reference in preference to Definitive XSLT and XPath.

The Gnome XSLT library is excellent, as is the xsltproc command line tool.

The Definitive XML Schema has not been opened - it would be nice to say that I'm savouring the opportunity to learn more about XSchema, but the limitations of ZSI would make this an academic exercise. My major gripe with ZSI is that it attempts to do too much, and does much of it badly - whilst it's possible to define recursive structures (e.g. trees) with XSchema, ZSI cannot handle this. The restriction to hierarchical data types is just too limiting.

Phil Armour writes a regular column in the Communications of the ACM which is well worth reading. I've been meaning to re-read these essays for a while, and apart from the odd dip, never really sat down to read them seriously. So to help structure this activity, I've prepared a 'reading guide' to the essays that have been published:

The essay are (in chronological order):

• The case for a new business model, August 2000
• The five orders of ignorance, October 2000
• The laws of software process, January 2001
• The business of software, March 2001
• Matching process to types of teams, July 2001
• Zeppelins and jet planes: a metaphor for modern software projects, October 2001
• The spiritual life of projects, January 2002
• The organism and mechanism of projects, May 2002
• Ten unmyths of project estimation, November 2002
• The reorg cycle, February 2003
• In the zone: the need for flexible roles, May 2003
• Closing the learning application gap, September 2003
• When executives code, January 2004
• Beware of counting LOC, March 2004
• Real work, necessary friction, optional chaos, June 2004
• Not-defect: the mature discipline of testing, October 2004
• The unconscious art of software testing, January 2005
• Project portfolios: organisational management of risk, March 2005
• Sarbanes-Oxley and software projects, June 2005
• To plan, two plans, September 2005

• The XML Revolution Technologies for the future Web is a great overview of XML and related technologies.
• This thread on comp.compilers discusses stack- and register-based virtual machines