ZDNET.com.au - Null Pointer

Get extensions going in Firefox, redux

edit@zdnet.com.au (Chris Duckett) — Tue, 24 Nov 2009 16:35:01 +1100

Previously on Null Pointer we looked at getting extensions working in Firefox betas. And that was fine until the fine folks at Firefox changed their minds.

This bug shows that Firefox now implements a per-application extension checking system. It's a little more fiddlier than before, but such is the life of a bleeding edge user. So once again we go back to the about:config page in Firefox and add the following variable:

Name	Value	Type
extensions.checkCompatibility.3.6b	false	boolean

Explanation for the variable's syntax is found at comment #18 of the bug.

On the previous blog post there was a comment that checkUpdateSecurity was not needed. I'd like to say that it is not needed, since removing security online is equivalent to dropping your pants and then attempting to get through airport customs, but when I re-enabled security, Greasemonkey refused to work.

I chose to re-enable security via a notification button in the add-ons window in Firefox, and that button changed extensions.checkCompatibility.3.6b from a boolean to a string — making the extensions stop working again.

The offending add-on notification that likes to reset variable types.

There is no nice graphical interface for changing a variable's type or to delete a variable. The answer is to go into your profile and update the user.js file.

To change the variable's type I had to find the line: user_pref("extensions.checkCompatibility.3.6b", "false"); and remove the quotes from around false.

And it is business as usual for Firefox functionality.

Comments (0) | Email this

Related Articles

Microsoft playing nicely with open source

edit@zdnet.com.au (Chris Duckett) — Tue, 17 Nov 2009 12:49:02 +1100

Before the Microsoft-centric developer world goes completely gaga at the company's Professional Developer Conference in the coming days, it's worth noting a couple of open source milestones that may have slipped under the radar.

Microsoft announced today that it is open-sourcing the .NET Micro Framework under an Apache 2.0 licence.

Peter Galli, a Microsoft open source community manager, said in the announcement that "Including the source code for almost all of the product also ensures that developers now also get access to the Base Class Libraries that were implemented for .NET Micro Framework and the CLR code itself."

Being an Apache licence means that, ironically, anyone can pick up the .NET Micro Framework and re-licence it under a proprietary licence. But not all the parts of the framework are being opened; the TCP/IP stack was made by a third party, EBSNet, and separate arrangements would need to be made with EBSNet for that source code.

Citing the use of the cryptography libraries beyond .NET Micro, Microsoft will not be releasing those libraries. "Customers who need to have access to the code in the cryptography functions will find that these libraries can be replaced," Microsoft program manager Colin Miller said.

Keep in mind that this isn't Microsoft opening up the holy of holies; the Micro Framework behaves in a different manner to the standard .NET Framework and has less capability than its .NET brethren due to the need for it to work on embedded devices.

If you are thinking of getting involved in the .NET Micro community, I'd give the website a little more time to bake.

But wait, there's more
Over the weekend, Microsoft also made its Windows 7 USB/DVD Download Tool available under the GPLv2. That's the same licence that Redmond has called "viral" for the longest time — and it's that same GPLv2 that made Microsoft release the code.

It began a couple of weeks ago when this blog post appeared and noted that some GPLed code seemed to be within the tool. The Redmond giant then pulled the tool from the Microsoft store and conducted an investigation of the claims. The claims turned out to be true — a contractor had included the library and Microsoft had failed to pick this up during code review.

To its credit, rather than damning the tool to program hell, Microsoft released the tool's source code under the GPLv2 licence.

While it is clear that this is not an action that Redmond wants to see repeated and is adjusting its code auditing procedures to pick this up in future, it shows a level of maturity in engaging with open source licences. They made a mistake, a licence violation, and were big enough to accept it and play nicely with the rules.

Would the Microsoft of 2002 accept its indiscretions and move on? I don't think so.

Comments (2) | Email this

Related Articles

Google open-sources JavaScript tools

edit@zdnet.com.au (Chris Duckett) — Mon, 09 Nov 2009 10:34:01 +1100

Google announced overnight the release and open-sourcing of a trio of tools designed to help JavaScript developers.

Dubbed Closure Tools, the search giant is continuing its push of JavaScript with a tool set including a JavaScript optimiser called Closure Compiler, a JavaScript library called Closure Library, and a templating system for JavaScript and Java entitled Closure Templates.

"The compiler removes dead code, then rewrites and minimises what's left so that it will run fast on browsers' JavaScript engines. The compiler also checks syntax, variable references, and types, and warns about other common JavaScript pitfalls," said the Closure Tools team in a blog post. "These checks and optimisations help you write apps that are less buggy and easier to maintain."

Closure Compiler results after optimising JQuery (click to here enlarge)
(Credit: Google)

The compiler is available via the command line, a RESTful API, or as a textbox on the Closure Inspector. Asking the compiler to optimise the popular JQuery library results in a 58 per cent reduction in file size but it comes with compilation warnings; making the compiler only remove whitespace gives a 40 per cent reduction with no compilation warnings — although it is outputting only compiler warnings and not errors, I don't promote blind faith in the output of this compiler without some testing first at this point in time.

It also seems that Google has fallen into the homophone hole again. Closure originated in computer science as a programming concept, with Lisp users already having the Lisp dialect called Clojure and Clozure as the common lisp implementation.

Comments (0) | Email this

Related Articles

Ubuntu can't cut geek support umbilical

edit@zdnet.com.au (Chris Duckett) — Fri, 30 Oct 2009 17:32:01 +1100

Ubuntu 9.10 Karmic Koala was officially released overnight and marked the eleventh release of the distribution. It's attractive, polished and measured, but fails "the grandma test".

When Ubuntu 9.04 arrived, ZDNet.com.au news editor Renai LeMay hailed it "as slick and beautiful as Mac OS X or Windows 7" and the distro has certainly not gone backwards since then.

In the following video is a CNET TV prizefight between Snow Leopard and Windows 7:

After watching this video, I believe that Ubuntu would be a contender if not the winner in this prizefight. It would certainly hold its own in interface stakes. Compiz (Ubuntu's enhanced graphical interface, which allows opaque windows) is the equal to Aero for Windows or Aqua for the Mac. In fact, making Compiz look and behave like either Aqua or Aero is a cinch. Ubuntu would hold its ground in stability too (it is a Linux distro after all) and would win in performance and compatibility — Ubuntu goes on hardware where Windows 7 and OS X fear to tread. It might suffer in unique features but would gain full points for value.

But if you follow the Ubuntu release trail, you are not going to be blown away by 9.10. It's a lot of 9.04 with another coat of varnish.

In a response to the photo gallery of Ubuntu 9.10 we did earlier in the week, one Twitter user wrote: "Ubuntu 9.10 Karmic Koala: 'Highlights ... include the addition of a new boot screen, an updated default theme...' Uhm, these are highlights?"

That is the reality of this release. It is not bursting with desktop transforming killer new features and applications, it's a release that has expanded to the cloud and the netbook, while focusing on fixing pain points for the desktop.

The One Hundred Paper Cuts project focused on "trivially fixable usability bugs that the average user would encounter on his/her first day of using a brand new installation", and although around only 50 bugs have been fixed, it does improve the desktop experience.

On the gloriously gushing side of improvements in this distribution is kernel mode-setting (KMS); if you have an Intel graphics chip then KMS is great. To have no more screen blinking on boot is fantastic, but this is how it should work. Non-Linux users though won't be converted with the "KMS means the boot to log-in to desktop sequence is seamless" argument.

When it comes to the new Ubuntu One service though, users of DropBox will feel that they have seen it all before; and at this time, they have a point. Beyond the integration with Tomboy, it behaves as a cloud storage service that is riddled with web server errors. The Ubuntu One site may say that it is beta — it is truly worthy of that status — but rolling it in with Ubuntu when it is clearly incomplete leaves a sour taste. Ubuntu One is not ready for prime-time and needs improvements now. To be fair I do expect that in the coming weeks it will come up to par.

However, the loop-jumping needed to set up synchronisation between Tomboy and Ubuntu One, getting contacts in Evolution, and synchronising Firefox bookmarks should not need tutorials that intense. Ubuntu controls the cloud service and the desktop clients, therefore it should be possible to have a Ubuntu One control panel with check-box selection to enable these services which would make it much easier for users.

Ubuntu One is but the first reason why I believe Ubuntu continues to fail the "my grandma can use Ubuntu" test. A lack of comprehensive testing of graphics drivers is another reason why grandma could not use Ubuntu without a Linux-using grandson on speed dial.

In the past week I've had the distinctly unpleasant experience of having to downgrade an Intel driver package. A Ubuntu 9.04 laptop had its Intel graphics drivers updated by Synaptic to a version that did not work with my laptop's chip — it simply made the screen black — no prompts, no errors, no feedback.

In this instance, downgrading the updated packages was all that was needed to fix the problems, but it highlights two cases of inadequate testing of critical parts of the distributions.

If Ubuntu truly wishes to take the Linux desktop to the masses, then it must be cautious in preserving the environment that it wishes for mainstream users to use. I'm certain that if I presented my Ubuntu laptop with an X server crashing to "technically-savvy" non-Ubuntu users, then they would have a hard time working out how to downgrade the driver, let alone the regular "ma and pa" users that Ubuntu wants.

Having the ability to revert the X server back to a VGA driver "safe mode" so that users could fix package errors in the familiar graphical environment would be preferential to having the X server spewing errors or simply making the screen black, forcing novice users onto the command line.

Until the computer shop down the road is able to fix these types of problems — which is unlikely — then grandma, mum, dad, and the entire family are still going to need to have a Ubuntu support umbilical cord to the nearest Linux guru.

Cut that umbilical cord and Ubuntu will be ready to go mainstream, and even though I think it could very well be better than Windows and OS X, the problems encountered when things go wrong still prevents Ubuntu from reaching its potential.

Comments (20) | Email this

Related Articles

Fixing Firefox 3.6's tab blunder

edit@zdnet.com.au (Chris Duckett) — Fri, 23 Oct 2009 16:10:02 +1100

I've been playing around with a beta build of Firefox's 3.6 browser for some time, and while it's been completely stable, its new tab behaviour has annoyed me.

When opening a link in a new tab, by default the tab will appear next to or near the tab in which the link was selected. Think Internet Explorer or Chrome tab behaviour.

Since IE, Chrome and now Firefox do this, it'll probably become de facto for browsers, but if you wish to remain a new-tab Luddite like me and restore Firefox's previous tab behaviour (where all new tabs appear on the far right of the tab bar), follow the following steps:

Type into the address bar about:config and hit enter.

The about:config page should look something like this.
Once there, enter into the filter browser.tabs.insertRelatedAfterCurrent
When you see the browser.tabs.insertRelatedAfterCurrent entry, double click on it to toggle the entry from true to false.

And you're done. Enjoy your new browser experience of knowing where to find the tabs you opened.

Comments (11) | Email this

Related Articles

How developers should ask for help

edit@zdnet.com.au (Chris Duckett) — Wed, 21 Oct 2009 12:00:02 +1100

Few people are better qualified than Tom Kyte to instruct developers on how to form questions asking for help.

Kyte is the man behind Oracle's long-running Ask Tom help desk — starting out as a magazine column and now an online forum.

In this video interview with ZDNet.com.au, Kyte explains how many questions he receives equate to "my car won't start".

"I'm going to do it to you — my car won't start, why? Give me the solution. I'm not going to tell you what errors I see on the dashboard; I'm not going to tell you if I filled the car up with gas recently; I'm not going to tell you if it's making a noise or not making a noise; but my car won't start, why not? There's insufficient data to even begin to answer a question like that."

Kyte also explains how the internet has changed how questions are asked, and how he doesn't reply with RTFM.

Comments (7) | Email this

Related Articles

What kind of company is Oracle?

edit@zdnet.com.au (Chris Duckett) — Thu, 15 Oct 2009 14:18:01 +1100

The formalities of Oracle's OpenWorld conference are done; Larry Ellison has delivered the last keynote of the event and covered a wide range of technology and topics. Yet after three days of conferencing, one question remains unanswered: what type of company is Oracle?

It's not the same database company from a decade ago. The company has clearly moved on and diversified beyond its core product. As witnessed by Ellison being able to deliver a multiple hour keynote without once mentioning the Oracle database product.

It could be a hardware company — once the Sun merger is approved. Ellison continued to take great delight in making jokes at IBM's expense, and reinforced many times the fact that he is willing to part with $10 million should anyone fail to double their performance on the new Sun/Oracle Exadata hardware.

Ellison humorously pointed that Big Blue has challenged his claim that Exadata is 16 times faster than its hardware — IBM says it is only six times faster, according to Ellison. "They might be right," said the Oracle CEO, and proceeded to take further delight in reiterating IBM's greater power consumption.

Oracle could easily be an enterprise resource planning/supply chain management company after Ellison dedicated the second half of his keynote to unveiling the company's new Fusion product. After a long wait, it's the result of a combination of technologies gained through acquisitions such as PeopleSoft, BEA and Siebel was on display. Even though Oracle will support legacy applications in this space for a decade, it is clear that Fusion is the path that Oracle wants its customers to traverse.

It could be that Oracle is partially a Linux company. According to a HP survey, 65 per cent of respondents said that they are running Oracle Enterprise Linux as their distribution for running Oracle Database (37 per cent said Red Hat with 15 per cent responding with Novell's SuSE Linux). BT has chosen to standardise on Oracle Enterprise Linux.

After years of acquisition action, Oracle has a software portfolio that few companies can match — and that is the power the company is starting to bring to the fore.

The Sun acquisition and new Fusion products allow Oracle to offer a complete stack for many enterprise needs. One of the themes for this conference was integration; however, what was not said was that while integration was possible with third-party products, it was naturally going to be easier and run better with Oracle products. Or to put it another way: all product integrations are equal, but some product integrations are more equal than others.

It's increasingly going to be hard to pigeonhole Oracle in any category as it now has so many fingers in so many pies. Had the Oracle and Sun merger been approved by regulators, it would have been interesting to see what McNealy and Ellison would have unveiled.

McNealy said in the opening keynote "I think Larry will like his new toy", and I think it is the observation to take from this conference. Oracle's database appears to have taken a back seat, with the excitement and attention of the company now focussed elsewhere, and is unlikely to ever return to the levels it once did.

Chris Duckett travelled to OpenWorld as a guest of Oracle.

Comments (0) | Email this

Related Articles

Non-professional Oracle wrestling

edit@zdnet.com.au (Chris Duckett) — Thu, 24 Sep 2009 12:06:02 +1000

The latest and greatest version of the Oracle database, 11g Release 2, was made available recently and as the resident technical person at ZDNet.com.au, it fell to me to take it for a spin. Little did I realise the hell that I had just walked into.

I will admit that two of the following mistakes are my fault entirely and could easily be fixed, but the third and final one is enough to drive anyone mad.

Mistake #1: choosing an "Enterprise Linux" as a test platform

The Asus server that was about to cause endless stress (see next mistake) came with a copy of Windows 2008 R2 on it. That's a bit of a problem since Oracle 11g Release 2 is only currently available for Linux.

No biggie, I'll just put Ubuntu server on there and we can live happily ever after. Wrong! The database's operating system requirements demand Asianux, SUSE Enterprise, Red Hat Enterprise, or Oracle Enterprise Linux and an X server. The headless Ubuntu server would not be any good in this trial.

Since I was testing an Oracle database, I figured that I'd take Oracle Enterprise Linux (OEL) 5.0 for a spin as well. Bad move. The catch with OEL is that you need a paid key to get updates from the online repositories, so you are stuck with the packages on the install DVD as a test. If any of the packages are buggy, you are stuck with them.

On the plus side, OEL had far less forms and registration loops than the other Oracle database-supported distributions, and clearly marked a single DVD download. I didn't want to download 8GB of Red Hat or SUSE only to find out that the second DVD is optional or more likely useless.

But in the interest of remaining on a supported OS, OEL was the best of a bad litter. I'm sure that CentOS or Fedora would work, but that doesn't get much love in the Oracle forums, as I saw people were told to get a supported OS as their first order of business.

Mistake #2: combining Linux and AHCI RAID

Since I am not a sysadmin everyday, I forgot about the whacky world of motherboard bus controllers and the varying degree of Linux support. Suffice to say that for the foreseeable future, I will not make this mistake again.

Installing Linux is not an activity I recommend with the Asus RS700-E6. Despite AHCI, RAID or legacy IDE modes being available, none were sufficient to install the OS and have it boot correctly without a significant amount of GRUB manipulation.

Legacy IDE mode was initially a non-event as it was dog slow, until an "enhanced" option was noticed in the BIOS and returned the drives to 21st century speeds.

But my favourite solution to this horrid situation was to use the Lenovo server with a proper hardware RAID card inside that worked with no dramas (hardware-wise at least).

Mistake #3: attempting to install Oracle

To the people that install, configure and maintain Oracle databases day in and day out: "More power to you!" I understand why it is that you are so well paid because if I had an option at this point, I would always have someone install, configure and create the database environment the way I wanted. I'd make the money back in the 10 years I would have lost from my life and save on dental fees I would have needed to replace my ground-away teeth.

The process for installing Oracle is arcane and should be better for a company that claims to be the leader in databases. Why do I have to go to the trouble of creating users, groups and environment variables by hand when other database and enterprise applications have been doing this for years? The answer escapes me; I'm hoping that there is a very good reason for this.

Over the course of wrestling with various instances of the Oracle install process I've had many things fail — so at least there is variety amongst the copious levels of frustration on offer — Java runtime crashes, kernel parameter errors due to not reading the instructions closely enough, and services failing during the configuration stage.

At least the database software itself always gets installed properly; it's the configuration of the listeners and the database itself that gives me grief. The configuration can be done manually, but wading through the documentation and ecma errors is a nightmare; not to mention that invoking netca causes the Java VM to crash hard.

It just shouldn't happen though. I'm not a 24/7 Oracle person, I'm a techie attempting to try the software out. If I was tasked with evaluating Oracle to run our next project on, I'd have to report back: "I'm sure it must be great when it is up and running, but once it is, pray to your deity of choice that it doesn't fall over or we'll have to get that expensive Oracle professional back."

So far it has taken days and I am yet to have a 100 per cent faultless installation experience, and I know that I could have MySQL, PostgreSQL or MS SQL up and running in hours by comparison.

For the time being, I am done with these vain attempts to get Oracle up and working. Perhaps I'll return to it when Windows' build appears — I can't say I'm looking forward to the next time our paths cross though.

Comments (5) | Email this

Related Articles

Microsoft creates open source foundation

edit@zdnet.com.au (Chris Duckett) — Fri, 11 Sep 2009 12:11:01 +1000

Microsoft has created the non-profit CodePlex Foundation to target increased communication between open source communities and software companies.

Citing an under representation of commercial software companies and their employees in open source, the CodePlex Foundation aims to work with particular projects to bridge the gap between open source and commercial worlds.

The Redmond giant has contributed US$1 million to the Foundation and filled out out its board and advisory panel with many Microsoft staffers.

Unlike other open source foundations, such as the Mozilla Foundation and GNOME Foundation, the Foundation said on its web site that it intends to address on the full spectrum on software projects.

Commentary

This is an unexpected and interesting move from Redmond. Don't think that this is completely like other open source foundations that you may be used to though.

Take this line from the Codeplex Foundation FAQ: "We wanted a foundation that addresses a full spectrum of software projects, and does so with the licensing and intellectual property needs of commercial software companies in mind."

Add to this that the About page states that companies will contribute code, not patents and that is what I think will stop the existing open source community from going anywhere near the CodePlex Foundation.

I can't see any patent-encumbered CodePlex project being accepted into, or contributing code into, any large existing open source project while still having the patent spectre looming overhead — it's something that the open source community has tried to avoid whenever possible.

But this is probably not that audience that the Foundation is aiming for — it's more likely to target purely Microsoft companies/developers and attempt to get them to open up a little. Allowing these companies to keep their patents will make it easier for them to engage in the Microsoft ecosystem but not in the wider open source world.

Comments (5) | Email this

Related Articles

No paper, no promotion

edit@zdnet.com.au (Chris Duckett) — Tue, 08 Sep 2009 14:12:01 +1000

It comes as no surprise to learn that HR people use IT certifications to choose between candidates when hiring, but in some organisations it can also inhibit career advancement.

"I think they limit themselves in terms of career progression, payment, and promotion," says David Hanrahan, general manager of Microsoft solutions at Dimension Data told me recently.

"It becomes a challenge to continue to promote someone who has a lot of experience and hasn't put in the hours and time that the rest of the people coming up through the business have to maintain their certification. It creates a level of disparity," he said.

Even if you have certifications, it is important to keep them up to date, especially as certifications change with the industry and increasingly move up the stack.

"TCP/IP, DNS and DHCP used to be part of the core networking exam and we would ask people [about them] in great detail [...] now we'll ask one or two questions to make sure they understand its implications, and move on." said Hanrahan. "We think certification is far more relevant in new and emerging technology areas, than traditional software and networking areas."

Getting onto the shortlist
The stronger your resume, the better your chances to get through the first round of interviews — it makes sense, but it was sobering to hear Hanrahan speak to the extent that certifications can make or break a shortlist selection.

"It's a differentiator — and we have to start to look for every differentiator we can when we are trying to get the best possible candidate" he said.

Hanrahan says that certification is a short-listing component, it's not going to be the hiring criteria; but it does actually make the job of hiring easier to use a baseline of industry-standard competencies.

If you are an industry veteran that has never taken the plunge into exams, you're not out of the hiring race, but to companies like Dimension Data, you will be starting from a handicap. "With the number of people in the market applying for jobs and particularly since we are targeted by people who want to come and work for us, we have a strong depth of candidates to choose from: they are at a disadvantage," he said.

Commentary
I have to say that I am guilty of being one of the "I'll take experience over certification any day" crowd, and never thought much of the general training courses. Being a certified software developer is never a sign that you are efficient, capable or intelligent — it's merely a sign that you can pass a test and cared enough to do so.

However, when you hit the realms of a level 5 certified Oracle DB administrator with a +2 Staff of 15 years' knowledge, that's when you know that this person knows a thing or eight about databases.

Hanrahan makes a very good point when he talks about using certifications for short-listing and mentioned that without qualifications, it is possible to miss out on how technically knowledgeable a candidate maybe in the first round of interviews.

I imagine the choice between an uncertified developer with 12 years experience and a developer of 9 years that has taken the time to get a couple certifications under their belt is a difficult one. Little wonder that Hanrahan places such a pragmatic emphasis on training.

That said though, I've worked with some really bad CCNAs that couldn't route their way out of a load balancer; but also some very good ones. When looking for that next job, you'll want every edge to standout and a course maybe that one thing that puts you over the top and gets you hired.

Comments (1) | Email this

Related Articles

Apache site hacked via SSH keys

edit@zdnet.com.au (Chris Duckett) — Mon, 31 Aug 2009 13:06:01 +1000

The Apache.org website suffered an intrusion over the weekend that resulted in the site being taken down.

The attack came via a third-party that was hosting the provider's servers used for hosting the ApacheCon site, according to a blog post by the Apache infrastructure team.

From that server the intruders were able to gain access to a backup server, on which they were able to create CGI scripts that were then automatically rsynced out to Apache's production web servers. Once on the production servers, the scripts were executed, spawning rogue processes which drew the infrastructure team's attention.

To the best of the team's knowledge, there has been no evidence that any downloads were affected and the intruders were unable to escalate their privileges.

Apache Software Foundation member, J Aaron Farr, said on reddit that the ApacheCon server would be rebuilt from scratch and that the team was still looking into how it was compromised.

Commentary
There's no need to run to the hills screaming and yelling in fear that the next iteration of Apache's web server could have "bad stuff" in it. This is an instance of defacing which only affected the Apache.org website itself. Farr said that Apache's svn servers were fine so the code is clean.

This is a good lesson on the potential downside of using SSH keys, which enable password-less log-ins for you, your scripts and anyone else that can get into your account.

While I am far from knowledgeable on Apache's infrastructure, it does seem a little concerning that the compromised backup server was the same server used to seed the production web servers.

Apache has said that it will come out with a full explanation once it has the whole story, where I hope it will announce the separation of the backup and seeding functions of the compromised backup server.

Comments (0) | Email this

Related Articles

Even the giants make mistakes

edit@zdnet.com.au (Chris Duckett) — Fri, 31 Jul 2009 16:05:01 +1000

Next time you begin to berate yourself for making that awfully stupid coding mistake remember that it happens to the best and biggest of us, including Microsoft.

A blog post on Microsoft's Security Development Lifecycle confirmed the cause of the latest exploit in Internet Explorer was due to a typo.

Here is the offending code:

__int64 cbSize;
hr = pStream->Read((void*) &cbSize, sizeof(cbSize), NULL);
BYTE *pbArray;
HRESULT hr = SafeArrayAccessData(psa, reinterpret_cast(&pbArray));
hr = pStream->Read((void*)&pbArray, (ULONG)cbSize, NULL);

And where is the typo you ask? It's on the pbArray's pointer, where the final line should be:

hr = pStream->Read((void*)pbArray, (ULONG)cbSize, NULL);

Despite all the effort that Microsoft puts into secure and proper code, its code analysis tools failed to detect the issue and the code reached production. The tools have now been updated to catch this problem and the developers will now have to use the new ATL libraries.

Now this would typically be the part where certain people start hooting and hollering about Microsoft's lameness and how they are personally the world's greatest coder.

However, I cannot don my asbestos suit for that. Humility precludes me, for I too, found a typo in my code from a 2006 article on JavaScript carousels. I have no idea how my clipboard contents, a URL in fact, found its way into the middle of some code and nothing was mentioned about it.

At least both typo-affected parties have learnt from this experience.

Happy sysadmin appreciation day
Today is also claimed as System Administrator Appreciation Day. The people behind this day say that you should shower your sysadmin in flowers and love for their unappreciated work — that's cool if your local sysadmin is a kind and personable fellow, and really, how many people like that are sysadmins?

If you have a BSMFH, I recommend broadcast pinging on your network all afternoon, endlessly copying many gigabytes of random files, secretly setting up an office MP3 server, and other general nuisances to make your sysadmin stop watching YouTube/reading your email like they do every Friday.

Of course, all these feelings for BSMFHs could be fixed if you would just give us all a bigger monitor and stop taking servers down during business hours.

Comments (3) | Email this

Related Articles

Microsoft Wave: By any other name please

edit@zdnet.com.au (Chris Duckett) — Tue, 28 Jul 2009 17:04:02 +1000

Picture yourself as a member of the Microsoft UK marketing team: you've probably busted your gut for the past few months putting together a new site and then the company in Mountain View goes and announces a product of the same name.

What do you do:

Brainstorm up an entirely new name?
Curse the gods and pick a new name?
Use up all your lifelines (50:50, phone a friend, ask the audience) and decide on a new name?
Nothing and hope no one will notice.

I swear that I never intended this blog to become a running commentary on Microsoft's marketing efforts, but when the software giant keeps selecting option D and shooting itself in the foot, it needs to be pointed out.

A showcase of Microsoft innovation is given an "innovative" name. (Click image to enlarge)

For a site that promotes Microsoft's world-changing blue monster, one would hope that the first thing to change is the name of this site. And as ZDNet UK editor Rupert Goodwins points out, what's with the copyright symbol in the logo itself?

A love of wave-esque names appears to be endemic to all arms of Redmond though, with a Microsoft Research project creating a device dubbed Wayve.

If you can explain the © symbol or anything else about Microsoft Wave, we'd love to hear from you in the talkback.

Comments (1) | Email this

Related Articles

PayPal launches Aussie developer program

edit@zdnet.com.au (Chris Duckett) — Thu, 09 Jul 2009 16:33:01 +1000

PayPal announced the opening of its certification program for Australian developers today, making Australia the first country outside of the US to offer certification.

A developer portal with all the documentation bells and whistles was launched in concert with the announcement, and currently contains sample code, forums and technical libraries for the 20 APIs.

Free developer days will be held between 20-24 July across the country by the company, with attendees picking up a voucher for certification, which would normally cost US$300. A free voucher is not a free laptop, but why look a gift horse in the mouth?

Comments (1) | Email this

Related Articles

Is Google asking for antitrust?

edit@zdnet.com.au (Chris Duckett) — Wed, 08 Jul 2009 16:43:02 +1000

Google has announced a new Chrome Operating System, designed for the web and with a browser baked directly into it — so much so that the entire OS is named after it. But the search giant should watch out: this decision seems designed to attract antitrust attention.

Imagine the furore if Microsoft announced Internet Explorer OS. Antitrust naysayers would drop from the rafters for the chance to lay the boot into Microsoft. Things have got to such a point that the next version of Windows is planned to ship without a browser in Europe.

What's good for the goose is good for the gander. If Opera has a problem with IE on Windows, they are positively going to blow many a gasket when they see what Google has in store. I'd expect the Norwegian browser maker to be drawing up a complaint to the European Commission as we speak — lest they'd want to be known as hypocritical.

Don't think for a minute that Google has not thought this through and does not have a game plan. At this early stage, I'd wager that one of the prime motivators behind the open sourcing of Chrome OS is to precisely avoid such potential antitrust claims. The argument would go that OEMs have the option to take the source and replace the baked-in Chrome with Opera, Firefox or WebKit.

That's the theory, but in practice, would an OEM actually do this? The Microsoft experience has told us no. Why endanger compatibility with Chrome OS's applications? As they are going to be mostly web-based, and presumably rely on the V8 JavaScript engine to provide performance, only a fool would swap out the fastest browser on the web.

There comes a time when companies move from being darlings to devils. Microsoft was once the underdog taking on IBM, Google was then the champion that would slay Microsoft, but today's announcement moves Google into new territory and new regulatory scrutiny.

If Microsoft is chased with torches and pitchforks for bundling a browser into an operating system, shouldn't the mob go berserk when a company the size of Google attempts to make the browser the OS?

Comments (4) | Email this

Related Articles

Microsoft misses the Outlook point

edit@zdnet.com.au (Chris Duckett) — Thu, 25 Jun 2009 11:40:01 +1000

Ask designers which mail program is the bane of their existence, and you'll find that Outlook tops the list. The reason why the most popular email reader is also the most painful is simple: it uses Word to render HTML emails — and Word does a very poor rendering job.

Continuing a decision made in 2007 to render HTML with Word in Outlook, Microsoft confirmed that Outlook 2010 will also use Word. In response to this decision, the fixoutlook.org campaign was created in an attempt to change Microsoft's mind.

Microsoft responded to the campaign with a blog post that sadly misses the point of the campaign. Instead of responding to the concerns of the Twitter folk, the post goes to lengths to explain the benefits of composing in Word.

As one poor sap that has had to create HTML email in the past, Outlook's treatment of HTML is a weight tied to the ankle of any designer or developer that steps into email. Lovers and overusers of divs and CSS need not apply as you will be returned to a world of tables and font tags that one previously tried to escape.

The Redmond giant correctly points out that an email marketing firm had set-up this campaign and that the Email Standards Project is not an industry body; however, the campaign has tapped into the latent issues surrounding Outlook and as Microsoft continues to hide behind the "Outlook plays well with Outlook" mantra, it will not win any friends.

Rather than move the state-of-the-art forward, Microsoft has decided to continue applying the handbrake and hopes you enjoy the stationary experience.

Comments (15) | Email this

Related Articles

IE8 marketing: MS takes the low road

edit@zdnet.com.au (Chris Duckett) — Fri, 19 Jun 2009 14:01:01 +1000

Take a quick tour around browser vendors' websites and you'll see them appealing to potential users with words like "passion", "difference", "speed" and "extensible". It's typical happy fluffy marketing speak.

Then there is Microsoft. The Internet Explorer 8 home page itself sprouts the usual marketing talk, but there is a promotion for browserforthebetter.com, a site where the gist is that if you download IE8, eight hungry Americans get a meal.

It's a good cause, it's a good idea — yet something didn't sit quite well with me. In retrospect it was the thought that I wasn't going to download and use IE8 anyway (especially as I have a Mac); yet I felt incredibly guilty not downloading IE8. I was faced with a choice: leave the page and have someone go hungry, or download a program I can't and will never run.

In the hopes that these donations are being based on download count and not on IE8 phoning home, I downloaded the file and then deleted it.

The whole experience felt dirty and wrong. I was being morally bribed into downloading an internet browser. The value pitch of the browser wasn't in improving an experience, it was emotionally blackmailing me and sending me on a guilt trip. But I couldn't put a finger on exactly what it was, it was like in The Castle and "the vibe".

Then I saw this argument on Twitter from a Microsoft employee: "Using FF or Chrome on Windows takes eight meals away from the needy."

Not the best way to express your company's message.
(Credit: Chris Duckett/ZDNet.com.au)

That comment snapped me from blackmailed to outraged. Read between the lines there and Microsoft is trying to tell the world that using a non-Microsoft browser is responsible for the ills of the world. It's not a good look, and does not reflect very well on Redmond at all.

With guilt-ridden blackmail taken care of, yesterday Microsoft moved to bribery and insults. Essentially a contest for US$10,000, the idea undoes itself with phrases such as "We've buried $10,000 somewhere on the internet ... you'll never find it using boring Safari. (So get rid of it, or get lost.)"

I chose to "get lost" and thoroughly enjoyed the experience of closing the browser tab. The readers over at Gizmodo came up with a list of insults for all the other browsers and state that overriding the user-agent string in the browser will fake the page out.

Taking another look at these experiences with IE8 promotions, they are both at their core, good ideas. Feeding the needy and giving users a chance to win cash are great ways to promote a product, but while the idea is fine, the execution is sorely lacking.

The contest is another attempt at Microsoft trying to look modern and talk to the cool kids, but failing abysmally and just making it harder for itself next time. There is being edgy, and then there is just plain insulting.

Similarly for the browserforthebetter site, it was simply distasteful and could have been done very successfully in a much better way. Microsoft's marketing department really needs to wake up to itself or get lost.

Comments (7) | Email this

Related Articles

If PHP then goto is the future

edit@zdnet.com.au (Chris Duckett) — Thu, 11 Jun 2009 12:18:01 +1000

Few things can spark more religious fervour amongst programmers than the mention of a goto statement. PHP has stepped into the middle of the firefight by announcing that it will be implementing goto functionality in version 5.3.

There is plenty of discussion surrounding PHP's decision, with PHP founder Rasmus Lerdorf defending the move, stating it is no different than breaking to a label and that it will not be an unlimited goto (it will not let jumps into loops or switches).

Despite Lerdorf's assurances, the wailing and moaning continues relentlessly. Why is this? To understand, one must realise that the vast majority of programmers have been taught that goto is considered harmful, is a bad idea and will bring pestilence to the land if used. Most students are happy to accept this, generally based on the experience of writing ridiculous spaghetti code in BASIC, so they move onto structured programming and eventually object-orientated programming.

This happens for very good reasons; a maturing programmer has more chance to get themselves in a bind by being unstructured and it is generally better for maintainability and understanding to avoid goto.

By going against the goto grain, PHP is challenging a tenet upon which the foundation of modern programming is built. The size of the challenge is only compounded by the fact that the change is occurring in version 5.3 of the language, a long time after the language muffed but eventually reached satisfactory attempts at implementing objects, and appears to be a step backward.

xkcd says it best on bad uses of goto (Click image to enlarge)
(Goto image by xkcd, CC2.5)

Here's a shock for the young, blindly-accepting-articles-of-faith type of programmer that has never had to use a goto: most high-level languages implement a form of goto, you just know them as break, switch and continue. And that most-revered of code projects, the Linux kernel, makes extensive use of the goto statement and is written in C — the most respected of languages.

What is good enough for Linus and the team of kernel hackers should be good enough for Rasmus and his clan of web developers.

It's true that goto is a dangerous weapon and in the hands of the inexperienced it's a devastatingly head-scratching weapon of spaghetti creation, but in the right hands it can remove the maze of nested branching structures nicely.

What does this all mean for PHP? Not much. Those programmers adverse to using goto will continue to avoid it at all costs (no one is forcing them to use it), quality programmers will be able to use it well and sparingly, and the bad programmers that create spaghetti will do so even more now that they can use the statement.

But they were going to create bad programs anyway, goto simply made it easier for them to get into a mess. They'll learn eventually.

Comments (15) | Email this

Related Articles

Flash tools hit beta

edit@zdnet.com.au (Chris Duckett) — Tue, 02 Jun 2009 13:57:01 +1000

The next iteration of Adobe's Flash tools have reached beta status and provide some concrete evidence of what Adobe was going on about with its prognostications of Thermo and changing workflow over the past year.

The problem with Flash Catalyst, the product formerly known as Thermo, was that thousands of words were not able to convey the interface, interaction or behaviour of an application in the way that four minutes of video can.

Now all those things Adobe previously said start to make a little more sense. There will no doubt be the argument that some developers like to do UIs their way and will hate the code generated out of Catalyst, but at a bare minimum I think any program that delivers some tangible code is far ahead of any mock-up that said developer would have received in the past — code is malleable, PSDs less so.

Adobe's Flex 4 framework, Flash Catalyst 4 and Flash Builder 4, previously known as Flex Builder, can be downloaded here.

Comments (0) | Email this

Related Articles

Google Wave to miss IE6

edit@zdnet.com.au (Chris Duckett) — Fri, 29 May 2009 12:58:01 +1000

Turns out that the "developer preview" of Google's latest creation, Google Wave, is not as open as one would expect, with the preview only being open to attendees of Google's I/O conference.

But there is another way to see it in action: as an attendee of Google's upcoming developer day conferences, one of which will be held in Sydney, where you will be able to gain access to the system before consumers get full access later in the year.

On the question of why not go live now with Google Wave, Jens and Lars Rasmussen cited that it is not ready for the general population. Wave itself is best viewed in WebKit-based browsers or Firefox with Opera support still coming together. For Internet Explorer, Lars Rasmussen said that there will be no IE6 support for Wave, but intends to target IE7 and above.

The reason for this is due to the lack of HTML 5 support for Internet Explorer 6, in particular, the use of web workers which are analogous to threading a web application.

For people that are worried that Google Wave is the next incarnation of Google Trapper Keeper, the servers that the communication is hosted upon do not necessarily have to be Google's. If you are interested in the protocol itself and are crazy enough to want to create an implementation of it, information can be found on the wave protocol site.

And if you want to try to explain what Google Wave is to others, so far I've managed to restrict the description down to a "hosted communication and decision engine".

I'm sure there is a more apt and succinct description for it, if you have an idea post it below in the talkback.

Comments (7) | Email this

Related Articles