La Tortue Cynique / The Cynical Turtle

Please, we're not in the Middle-Age

I've just read Meg's Dial-Up Revelations at O'Reilly, and some things annoyed me... First, it's not pain au chocolate but pain au chocolat. (The only point in this first remark is to prevent people from reading further, they'll just say "those french are so arrogant" and leave, goodbye.)

The real thing that annoyed me is the correlation made between latest technology and standards vs. efficiency in data transfer (thus the dial-up). Does a RSS feed give me more than I had more than 10 years ago with Gopher? It's just a sacralization of the "XML over HTTP" couple, the true magical "one size fits all" technology of the future... And for the CSS, from where did come the bloat in HTML that CSS remove? From Netscape right, the guys that today sell us `the standard, nothing more, nothing less`. Isn't it ironic?

(Why are my writing always more harsh than my thoughts?) What I know is that the web is still young, the standards of today are going to change dramatically within 10 years, and that I really don't know what it'll look like in 30 years. As such, XML as the unique foundation for all future communications in our i11l world doesn't convince me, and we should go on moving forward, and not be stopped by standards. Just regarding application protocols, today are still alive NNTP, IRC, SMTP, FTP, HTTP... many have faded away, some are still not there: instant messenging, broadcasting, remote procedure call... It takes time to settle down, IPv6 is going to bring a new field of challenge and opportunities for instance.

To end up this messy post, I'll just precise that I'm a DSL user in France (not an early adopter), we've got electrity, heating and even television. We don't like to eat in a rush, and when we are, it's just a question of asking for the last meal and the check at the same time, or just go up an pay while quitting (Cyberpunk, 2002/12/21 07:29) lien permanent

A vision of hell

http://freeroller.net/page/shemnon/20021220#eclipse_does_swing(Cyberpunk, 2002/12/20 15:55) lien permanent

Eclipse in Swing?

Russ would like to see Eclipse coded in Swing... Please no. Eclipse is snappier on my machine than my 60kB small programs in Swing...(Cyberpunk, 2002/12/18 17:18) lien permanent

Back on the Wired bias

The "journalist" article on Chandler is online. Even if the OSA Foundation website refers many times to Outlook, it's a little convenient for the Wired "journalist" to forget the role Mitch Kapor had with Lotus Notes. Should we see Chandler as an opponent to Outlook or as a successor of Notes? The first shall not be dismissed, but I feel the second brings more sense in this affair.(Cyberpunk, 2002/12/15 10:05) lien permanent

"Wired" fakes Microsoft's bashing

Mitch Kapor has been deceived answering some Wired "journalist".(Cyberpunk, 2002/12/14 15:37) lien permanent

Carlos Perez's rant

on JDO. Reducing this log entry to a rant is not fair, because it's thought. Everytime I read something about O/R mapping, I can't help it, I remember one project where we had CMP beans generated with Visual Age's Persistance Builder... whose links were affectionately destroyed by hand, because the object's graph handling was so poor.(Cyberpunk, 2002/12/13 16:32) lien permanent

Back on Word...

Andy Oliver replies that he likes Office 97 and that it's reliable under Wine. I'm not that against Word, I think I started word processing with Word 2.0c under Windows 3.1, it was quite fun After some years, I'm just a bit tired of how non-productive I am with Word. Typically, it loses my numbering after 15 minutes of work, I don't like the way it selects text, I don't like to be forced to crawl the network to find the right stylesheet.

There was a time when I was doing my editing with LATEX, on a 486-SX40 with Emacs, and previsualisation with xdvi with WM2 as a window manager... and that was simply great. Editing a plain text file is quick, saving doesn't take hours, revision control was a breeze and the typesetting is just gorgeous. The only really painful point was to insert graphics, and more, drawing on xfig with a 800x600 resolution was a little difficult, because the controls didn't fit! That was some years ago...

So when after 1 or 2 hours, Word on NT4 crashes for the second times (Lord knows why, not even a single graphic in that 10 pages paper), I'm a bit upset, and I rant (Cyberpunk, 2002/12/13 16:01) lien permanent

Why programmers don't like to write documentation?

Because they are given tools like this one:

Une exception d'application s'est produite :
        App : WinWord.dbg (pid=293)
        Quand : 11/20/2002 @ 15:26:2.408
        Numéro d'exception : c0000005 (violation d'accès)

fonction : HidOfDlg
        30808d23 8b542404         mov     edx,[esp+0x4]          ss:0102d5e3=????????
        30808d27 8b01             mov     eax,[ecx]              ds:309df6d8=309e3bd0
        30808d29 56               push    esi
        30808d2a 395008           cmp     [eax+0x8],edx          ds:318e25d6=????????
        30808d2d 7504             jnz     HidOfDlg+0x49 (30808d33)
        30808d2f 5e               pop     esi
        30808d30 c20400           ret     0x4
        30808d33 85d2             test    edx,edx
        30808d35 895008           mov     [eax+0x8],edx          ds:318e25d6=????????
        30808d38 7426             jz      HidOfDlg+0x76 (30808d60)
FAUTE ->30808d3a 8b12             mov     edx,[edx]              ds:0000001e=????????

No more Visual Basic after tomorrow

For one month my job is to maintain a client-server application in VB5. I has not been the most fun thing in my life, but at least it provides some occupation. Quick pros and cons of VB5 :

Pros

Easy interface designing

Keyword With enhances readability for multiples operations on an object

Very well integrated debugging

Cons

Fixed layout, resizing must be handled manually

No always well-suited events

Poor string handling capabilities

Poor readability due to default properties

No inheritance

Random crashes of the IDE

Deployment hell (why a simple demo app, with no dependencies doesn't run on others 'puters?)

No 'Inc' function (this function is one of the first I used proudly in Basic)

Overall it has not been a bad experience, and it leads to a little wishlist for Java.

Syntactic Sugar

A 'With' keyword

Behavioural change

A true static resource oriented user interface definition. Two years and a half ago I designed my own based on XML. The performances were quite acceptable, there is so much resource used with Swing that you can add many IO and parsing before you feel the overhead. I don't know how it would perform today (proprietary source code, and I don't work there anymore). Of course it depends very much on the graphic toolkit you use (AWT, Swing, SWT, QtJava, Java/GTK, dog.gui...), that's probably why there are multiple ways to do it (serialization, LUXOR, code generation...) and no one really as clean as a good ol' resource file compiled.

(Cyberpunk, 2002/12/13 07:54) lien permanent

Sans titre

Why companies lock us with tools as unreliable as Microsoft Word?(Cyberpunk, 2002/12/13 04:09) lien permanent

From LGPL to ASL

Tapestry eventually changed its license from LGPL to ASL. License was probably the main restriction for Tapestry from joining the Jakarta project. Tapestry is a direct concurrent of Struts, and overlaps with other Jakarta projects. Tapestry is bundled with Jetty, a direct concurrent of Jakarta Tomcat... Let's see if Tapestry joins the Apache Software Foundation and if it mixes well...(Cyberpunk, 2002/12/12 16:30) lien permanent

Information decay

Returning on the hungarian notation for Java, one very important thing to consider is how to maintain coherent information. If you don't pay attention to maintain your "fbsStatus" and other "hcnChilds" up to date with their real type and use across the refactoring cycles, it's more a danger than a help. That means we must have a dimension of variable naming in the refactoring patterns.

Let's take a simple example. A property "available" has a boolean type and its no more sufficient for the exemple use, so you change its type to int.

boolean m_bAvailable => int m_bAvailable

In parallel, you must rename it.

int m_bAvailable => int m_nAvailable

The type has changed, you must change the accessors.

void setAvailable(boolean p_bAvailable) => void setAvailable(int p_nAvailable)
boolean isAvailable() => int getAvailable()

And then change the calling code, that is change the variable types and names where applicable. The complexity of this refactoring is almost doubled compared with the one without hungarian notation. So we need style checker which can handle this (I wonder if Checkstyle does it?). By the way, there is a little problem now, "getAvailable" doesn't sound right at all, it should be "getAvailability", *sig*. (Cyberpunk, 2002/12/03 05:38) lien permanent

I for interfaces?

Some days ago on Otaku, Cedric's weblog, "I is for Interface":

The standard argument against such a practice is, again, that it breaks encapsulation: "I am dealing with a type, I don't care if it's an interface or a concrete type."

It sounds good, we don't do such discrimination in real life with peoples, animals and stones for instance.

Well, you should, because they are not equivalent. For example, you cannot "new" an interface, and it's the kind of information I would like to have right off the bat

This is the same for abstract classes, singleton, most of the utility classes, but we don't use AMailMessage, SORB or UStrings. Of course it's an important information, but should we rely only on naming schemes to provide us the vital documentation? An API, a product, is more than just a set of interfaces and classes, it comes with a vision for solving a problem. You don't get that vision with only names, and once you get the vision, I think you don't really need this "I".

There is no harm in distinctly flagging your interfaces with an "I", I believe it makes the code much easier to read because you can now see clearly which part of it is reusable and adaptable as opposed to your own private implementation.

Yet there is in fact no real harm, refactoring tools can handle nicely, even if would like to be able to put easily a comment everywhere the refactoring tool let its footstep. And such a naming scheme is easier to enforce than a good documentation. As for seeing which part is reusable and adaptable, I've seen a product where all the classes from the public API were prefixed with "API", that made the documentation unbrowsable quickly.(Cyberpunk, 2002/12/02 02:06) lien permanent