Apology? or are we starting the blaming game.

First, we have Daniel Lyons using Forbes “follow-through” tradition as a shield to say that he was “snowed by SCO“, now TG Daily carries a similar “I was lied to too” article by Rob Enderle. Were they forced to do it by their editor? Or were they trying to salvage their reputation? I don’t know and quite frankly, don’t care.

Enderle choose to put the blame partly on hot-headed internet users who “push him to SCO”. Was that necessary? With so many internet users we have hotheads for Microsoft, hotheads for Linux, hotheads for Google, hotheads for just about anything, with exception of “hotheads for SCO” of course. It felt like someone trying to defend his decision. I must say, although I wish I am at a higher plane than him when it comes to forgiving my enemies, I am not sure whether I will refrain from attacking my enemy the way he did. I do sincerely hope I can just swallow that pride and extend the olive branch just a little further.

Enderle also choose to hide behind “I did not say that, they twisted my word”. Unfortunately, I do not buy that his word were twisted by others. In communications it is not what one intends to say that is important, but what is received that is the important part. I learned this the hard way, but that is an invaluable lesson.

Compared to Enderle, Lyons’ was a better written piece. He did not try to pin any blame on his opponent. In fact, he went further by acknowledging some of his opponents got it right. Kudo to him for that!

Both mentioned Groklaw. Groklaw was never kind to both of them (and two other ladies for that matter). Unfortunately my view is that Groklaw successfully point out the fallacy of their arguments and like Groklaw, they both have something to sell.

Bottomline, why do one pay for analyst’s opinion? Do we want pay for them regurgitating PR statments from company, or do we want them to use their brains to check their sources, digest the information, and make informed decisions? Do we want our analysts to stay cool under pressure, especially random noises from the internet that “threaten” them, or go against them with a vendatta? Do we want our analysts, when spotting something is not it seems it is, to tell us so, or continue the trend until the very last minute when it all tumbled down? Or do we want our analyst to use one event to advance his attack on another, larger attack on another event, AFTER he got “smart” about the first event, but before the final curtain for the first event?

Gnumeric’s view on OOXML

Let’s face it, I am not writing a office application, neither is 99.9% of the people discussing this topic. Hence, it is refreshing to see the viewpoint of an Gnumeric spreadsheet writer.

His view was a nice insights into the actual work of implementing OOXML and ODF. To me, it highlighted another hurdle in implementing ODF: Most office applications are written with EXCEL compatibility in mind. Meaning it is easier to implement OOXML because OOXML reflects XLS structure closely. Part of the difficulty in getting a software suite running, any software suite for that matter, is to create sufficient software “foundation” component in the suite to support various functions expected by the software. It’s time consuming and difficult job, but once that is done, adding new things is a matter of building on top of them becomes a relatively simple task.  In this case OpenXML enjoys the advantage that the “foundation” components were already written and it simply had to ride on them. Of course, the fact that OpenXML is more “machine-language like” than full structural XML helps. On the other hand, any one writing ODF components probably have to twist and hack to fit into existing, XLS-centric code.

In my opinion both sides have a valid claim. It can be manipulated by both sides. Pro OOXML people will rejoys in Jody’s view that OOXML is faster to implement, coz they can reuse existing knowledge on reverse-engineering binary XLS format used by Excel. OOXML opponent will say that is only to be expected, if you write your application to reflect XLS.

It is unfortunate that both parties had decided to fight on these issues as the author presented it. If we measure it in terms of speed of implementation, I think OOXML is going to benefit from large existing implementation that is geared towards understanding its predecessor format. To win this argument, Pro ODF people has to escalate beyond implementing OOXML or ODF. I would like to see Jody, once the Gnumeric team completed the OOXML and ODF work, comes back and tell us which one they prefers, as a long term solution to file format.

less work better, or follow convention better?

As a programmer, I find this an eternal struggle. My comment to BSI on OOXML for example, listed the following “problems”

• Too many *.rels file. Easier if you put them in one big  file call MANIFEST/manifest.xml
• rPr attributes makes the WordProcessingML unnecessarily long
• I need to read  .rels to find out which file contains the main document, rather than just reading contents.xml as in ODF. I believe in the end, a lot of small developers will simply assume (thus requiring everyone) to follow the same convention as MSOffice.
• Should had reused html colour scheme instead of defining its own
• Should had reused existing standards
• VML, being deprecated, have no place in a new standard

And also I dislike the fact that SpreadSheetML uses shared strings.

I still believe a few of them, VML, reuse existing standard, colour scheme, are justified. The rest (too many *.rels file, rPr,  MSOffice Convention
and share string), which I classified as burdensome for small developers. Not surprisingly, BSI did not accept the “small developer” arguments, especially since I phrase it as “a lot of (unnecessary) work to do just to get a piece of data out”. Since I wrote the comments, I had been plagued by the thoughts that I am simply too lazy.

I am glad that I am not alone in thinking less is beautiful. On Jason Matusow’s blog, a user by the handle Simone commented that one reason he favoured OOXML is that it allows custom schema, which is easier to do than the XForm approach in ODF. I always thought that this is true and is one thing that can persuade developers not working on office productivity application to favour OOXML over ODF.

On the other hand, some claims by ODF people that XForms has its merits are true as well. Encoding your own XML data in XForm makes it easier for ALL to understand and manipulate the data.  The other, custom schema can leads to “proprietary version” of OOXML, i.e., people embedding data in schema to ensure nobody but a particular brand of OOXML-conforming software can read it, while is a valid argument, will probably be a plus for most developers, especially those not interested in sharing data beyond MSOffice.

Over the years, I had find that taking the easiest way out is never the better solution if a standard conforming way is available. It is true that the initial cost of developing the program is higher because the developers have to acquire new skills and the setup is more difficult. For most non-trivial program that I wrote, and especially true for programs that survives a few years or more, I do find that sticking to standards are the best way.

So why is custom schema useful? Easy to read/parse since one is in complete control. On the other hand, if you take the trouble and write it in XForm you will likely find that a minimal rework is needed if you need the same thing for other purpose. I always says that one have to seriously consider taking the standard approach if you can see that others more informed than you takes the trouble to create a standard and that standard is around for a good few years.

Is OOXML custom schema ability attract developers? You bet! Is it a better way than XForms? I simply do not know: Insufficient data. In 10 years time we  will probably STILL do not know the answer. Why? If the market is going ODF/OOXML this argument is moot. If the market is going to be fragmented by both ODF and OOXML we will still be having this debate.

Lotus Symphony: First Impression

As you all know, IBM just released Lotus Symphony, a free Office Application for Windows and Linux (with Mac to follow). I downloaded the Windows version and took it for a spin. Firstly, as a programmer, I am naturally curious on how IBM used Eclipse in Lotus Workspace. Not wanting to spend a penny on Lotus Workspace, peeking inside Lotus Symphony will simply have to do 😉 . Second, the screenshots look good and I want to compare it with Microsoft Office, OpenOffice.org and iWork.

First thing is to install it. The download is a self-extracting zip file, which you have to unzip then click setup.exe. Like any Windows-based software, it is easy. It is also less confusing than OpenOffice.org’s installation. Comparatively it takes much less time to install than Microsoft Office and OpenOffice.org. The only surprise is the prompt to “restart” Window, especially if you are used to Eclipse. I am sure there is a good reason.

Then come the first disappointment: When you click Start->All Programs from Windows, you find that it actually installed three icons, one each to “Documents”/”Presentation”/”Spreadsheet” instead of putting them into a folder named “Lotus Symphony”. This is dissappointing because it is an old practice whcih Microsoft abandoned with Office 2003 and above (thank god). The second disappointing thing is the automatic installation of three shortcut buttons for the three applications on my desktop. I would had prefer the installation process ask me whether I want this done rather than assuming I want it. Of course, these are only minor issues.

Starting “Lotus Symphony Documents” reveals the main UI. It may not be as flashy as iWork, but it is certainly a refreshing change from Microsoft Office. It use a right hand panel which works a bit like iWorks right hand panel. It may be more simple than iWorks, but it is for the better coz it is more easy to use than iWork. Both iWorks’ and Symphony’s right hand panel performs virtually the same function as Microsoft much touted “Ribbon System”, but being placed on the right hand side rather than the top, it suits the current cinema-screen style monitor standard better. The “ribbon system” had to be made “smaller” so as to expose more screen area for the main document, but with right hand panels this problem does not exists, and it has more space to make the UI in it “larger” and better arranged.

What I really like is the “documents” tab where all your open documents are listed there. This means rather than openning separate application for spreadsheet, document and presentation, you have them all in one window, your main window. Back in the 90s when SUN release the first version of StarOffice, it has something call the SUN Desktop which contain all your open documents and actually do more-or-less what your main window do. However, it was so awkward and get in the way of your work that it was a relief when later versions of StarOffice/OpenOffice.org dumped it. This implementation by IBM actually helps you in your work. I also like the “thumbnail” feature (Windows->Show Thumbnails or Ctrl-T). Also note that your main menu bar, the one with “file menu” etc on it changed with different type of document. Well done IBM!

In terms of first impression on usability, it is probably better than Microsoft or OpenOffice.org. To be honest, I actually prefer its layout than iWorks: it is as clear, effective and well positioned as iWorks, but also simpler to understand.

Compared to OpenOffice.org, it is better thought out. Compared to Microsoft Office, it is less intimidating and less gimmicky. Compared to iWorks it has better functionality and Lotus Symphony’s simplicity stands out.

When it comes to website design, Symphony’s is less flash than OpenOffice.org, Microsoft Office and iWorks and wait for this… focused on the user! This might sounds surprising but if you look carefully, OpenOffice.org concentrated on developers; Microsoft Office concentrated on getting you to navigate the complex and different configuration of MSOffice landscape; With Apples, we find the classic Apple clear and simple presentation of what iWorks is, and is therefore simply a promotion page for the product. Perhaps this is because the website is still in its infancy. We just have to wait and see.

What Lotus Symphony lacks is what OpenOffice.org lacks: Nice templates and clip arts. At least with Lotus Symphony, it is easy to find ClipArts on its website. OpenOffice.org’s is buried somewhere inside the website.

Overall, Lotus Symphony is a credible challenger to MS Office. It does not have the “half-finished” feel of other office software, and compared to OpenOffice.org, the UI is more appealing. Eventually, the question of whether you will fork out about 100 dollars for Microsoft Office (Student Edition. Let’s face it, most of the world quality for this discount) will boils down to whether is the templates and cliparts collection in Microsoft Office is worth the money. At present, I say yes. Thus, there is no surprise that I think Lotus Symphony (and OpenOffice.org) should work on templates and clip arts to appeal more to end users.

From a programmers’ viewpoint, the whole experience have Eclipse RCP written all over it. The progress screen during installation shows standard Java reverse URL style (com.ibm….) and the word (org.eclipse…) appears a few time. The splash screen has the “built on eclipse” logo and the UI itself have several unmistakable features that shows it coming from Eclipse, e.g., Help->About IBM Lotus Symphony screen looks eclipse-like and (wait for it…) has “Plugin Details” and “Configuration Details” button. I also have a peek at the installation directory. Organization-wise it is a bit confusing intially, but an Eclipse programmer can easily decipher it. The difference might have to do with programming nicety such as separating source code with different licensing terms to ensure no accidental mixing and the ability to revert to old configuration quickly. One surprising thing is I cannot find any icons with “Documents”/”Presentation”/”Spreadsheets” to click on to launch the application. Eclipse RCP developer will not find this surprising, coz we know our favorite “<myapps>.exe” is simply “eclipse.exe”. Nevertheless, it would had been good to have these dummy “Documents.exe”/”Presentations.exe”/”Spreadsheets.exe” if only to make attaching them to “Start” menu easier for joe user. By spending 5 minutes inside the Installation directory, my appreciation of Eclipse actually improved!

Eclipse RCP programmers will also appreciate and know how IBM deliver functionality such as the top document tabs etc.

While OOXML vs ODF saga receive the most attention now, there is another front where Microsoft is targetted. Eclipse developers like me know that the Eclipse Foundation is also trying to wow developers who needs to “relearn” Windows API (for Vista) to use Eclipse Platform (RCP in particular) instead. Lotus Symphony will be a good showcase and is a convincing example. I am sure it will win more than a few converts in time to come.

Office Applications had never been so sexy since early 90s

IBM announced a free office productivity suite known as “Lotus Symphony”. Never had office applications becoming so sexy since 1990. For those of us who were old enough to be around, we don’t really need The Wall Street Journal’s gentle reminder but won’t mind it, we will quickly conjure up the image of the Office Productivity Application war at the time.

Although several Office Productivity applications are available, mainly open source or beaten rivals, for a long time now the only two darlings are Microsoft Office and OpenOffice.org. Lately we see web-based office productivity tools, and a lot of people probably thought desktop Office productivity tools days are numbered. Now come Lotus Symphony, which certainly spice up the whole market.

IBM obviously aimed Lotus Symphony at Microsoft Office users. The target price is fantastics, compared to Microsoft’s cheapest version of Office (The USD3 bundle for China does not count). From David Berlind’s account of Microsoft PR activity, they appear rattled.

Is this yet-another-attempt to detrone Microsoft? Realistically speaking, it is about chipping away Microsoft’s dominance in Office Productivity Suite. And that is whats that count coz it is about chipping away the monopolistic power that Microsoft enjoy. At the mean time, if you have any doubt that this is another weapon to be used in the ODF vs OOXML battle, just go through the tour for the Word Processing apps.

Interesting …

Its refreshing to see that Microsoft is not the only one trying to game the system. OpenMalaysia found this old document that claim Sun did it with Java, with Microsoft crying foul. Interesting reversal of fortune.

The good news is, if SUN did game the system, it did not succeed. Currently it is too early to say Microsoft will not but if history is an indicator…

The significance of losing the Trustee in Microsoft EC antitrust appeal

On the surface, it appears that Microsoft loss seriously on its appeal of EC antitrust. Yes, it lost most of its claims bar one, the appointment of the trustee. That came as a surprise to me, an outsider.

The remedy used by EC to date is two fold: One is the punitive fine levied on Microsoft, especially those that are still being incurred on a day-by-day basis until Microsoft complied. The second is the trustee. EC appears to be using the punitive fine as a means to tighten the noose and is not the main tool to enforce compliance. The trustee is therefore the main tool to assess compliance. What the court says is that EC has no legal power require the appointment of the trustee and to get Microsoft to pay for the trustee. Hence, the EC either have to find some other mechanism, or enter into a “contract” with Microsoft with reference to the trustee.

If I were a betting man, I will bet on Microsoft dragging this issue on how to assess compliance as long as it can. Simultaneously, it will appeal this appeal. Time is on Microsoft’s side.

As such, I do not think that the judgement, in reality, is a lopsided against Microsoft.

Nonetheless, it is a significant victory for EC because it upheld all its decisions. The trustee issue is an annoyance and hey, one do not really expect the status quo to stay on appeal.

Really unfortunate that SCO filed for Bankruptcy

Well, the news is well and truely out, SCO filed for bankruptcy. Not a surprise for all of us who is following the SCO saga at Groklaw.

This act by SCO automatically postpone the SCO vs Novell trial tomorrow. Given that SCO files for bankruptcy protection the Thursday before the trial on Monday, I cannot help but think that SCO did it to maximize inconvenience to all others. Novell’s lawyers’ now have to abandon trial preparation, prepare their response to this event, then ponder what to do for the next week where they had pencilled in the trial. Novell’s will still have to pay for the trial preparation done by its lawyer. All those good folks volunteering for Groklaw now have to cancel their travel arrangement and try to get a refund. Judge Kimball will find that he does not have a court hearing to attempt. Mind you, he probably can find solence in catching up on his workload. So does Novell’s lawyers. The only people that are least inconveniented seems to be the SCO team. And finally … Novell’s is unlikely to see the the money it said it is owned from SCO’s SUN and Microsoft agreement and which  will be determined at trial.

As for my part, the main battle, Unix Intellectual Property claim, was decided a while ago. Whatever remained of the trial are simply tidying up the case. However, it still looks as if SCO manage to escape final judgement for its evil did. In a sense, US legal system worked. It is just extremely costly to find out whether it works or not.

When SCO launched its case, people, including me, thinks that it is merely hoping to be bought out by someone, probably IBM.  IBM, to its credit, did not budge. Then we see SCO’s share going up and down, creating opportunity for some opportunists (lets just call them vultures) to cash in its misery.

As for SCO itself, it used the lawsuit (which I am sure it knows it cannot win) to buy time to create Mobile Linux, then use bankruptcy to ditch its debts. If the company probably survive this bankruptcy, and Mobile Linux will be its best chance to survive. Ruthless thinking, of course, but a good one.
Finally, the name SCO is in the mud now, so I would not be surprised to see it renamed to something else, probably ME Inc, if SCO emerge from bankruptcy.

Windows Update should NOT update itself, or the update list if I choose “NOT to update at all”. Period

Nate Clinton of Microsoft responded to accusation that Windows Update “updates itself” even if you choose not to update. He asked for comments, and he shall have mine.

My position is if I choose to disableWindows Update completely, then I means hands-off my windows installation. This means NO communication with the Windows Update server whatsoever, no modification on my windows installation, i.e., no self-update and no update list updates either.

All these other thing that Clinton choose to justify Window Update self-update are related, but irrelevant. Clinton is talking about managing Window Update  if you elected to choose to use it. I will say that if I choose to be notified but not automatic installation, I do NOT want Windows Update to update itself. Find another strategy to manage Windows Update without self-update.

Bottom line: You ask me whether I want to “disable”, I said yes. With this, I elected not to even want to know whether an update is available.

Data Lock Up hypocrasy and Independent Implementation

When I check Jason Matusow’s blog today, I find two interesting post, one about Data lockup in Web 2.0, and another about Independent Implementation of OOXML. Although Matusow did not link both together, lets do this as they have significant overlap.

I am glad to hear that in the first article, Matusow is advocating control over his own data rather than letting web 2.0 vendor control it. One way of doing this is to insist the vendor make the source code of the web apps available for other to implement it. In this case we do not really need any standards, just the unconstrained rights to run the web apps. On the other hand we can see the issue of control over data as the ability to take one’s data away from the closed proprietary system of one vendor to that of any vendor. This can be done if both vendors allow me to import/export data  a common standard. Although useful, that common standard need not be open, just implementable by multiple vendors. This neatly links the first article to the second.

Rule of  thumb says that the longer and larger a  standard is, the more difficult it is to implement. Conversely, a shorter standard makes life difficult for implementers as a results of lack of necessary information. One thing a good standard is to find a strike a balance. I believe OOXML size, at 6000+ pages is too long, badly written and shows a lot of implementation details that are unique to a product and should not be in the specification. Matusow is arguing that Part 4 of the standard is long because it gives implementer the chance of commercial grade implementation. If length is the key, then may I suggest Microsoft simply make the whole Office source code available for everyone to read, since OOXML’s own design goal is full compatibility with Microsoft Office document.

As for bad editorial control, it affects the quality of standard and should not be taken as lightly as Matusow did when he argues that these are fixable and imply that it will be fixed in the run up to Feb 08. Bad editorial control itself is capable of sinking a standard for a good reason. It creates confusion. A good standard allows its reader to read a small section and get all the information it needs for implementation. With OOXML as it currently stands, even with the section working on WordProcessingML alone, one must read the whole complete WordProcessingML section to work out (infers) how to implement it since one have to gather the information from a few places. It is a recipe for confusion if one have to gather information than infers the meaning.

Let’s completing the circle and go back to the idea of Data Control: One can argue that a hypocrisy that Matusow choose to ignore is not the issue its employee’s role in data lock-in. It is obvious that the reason MS is keen on OOXML is that a rival standard, ODF, threaten the data control ability of Microsoft’s own offering. If it were not for the shock of Massachusetts dumping MS’s own OOXML predecessor we probably will not have this discussion today.