Friday, November 10, 2006

Week-end bits and bytes...

I was browsing Laurent Schneider's blog and he pointed me to co.mments - a way to monitor comments left on blogs, even if the blog does not support such a thing (like blogger doesn't).  Looks very promising, that would always drive me crazy - when I left a comment, I'd usually want to see what others had to say about it later.  This will let me do that easily now.

I don't read Joel on Software often, but this entry on measuring "software productivity" popped up and caught my attention.  I know exactly what he means about "scamming the system".  I was talking with a friend recently who works in an organization where metric based performance is practiced.  The "best of the best" in the organization is chosen not by who sells the most, who delivers the best value, who is really good - but rather by who did the most "checklist items".  The goal was to make performance management "entirely objective" - make some things worth "points" (like Joel talking about 'function points').  The end result - the best performers end at the bottom of this measurement process and the ones that do not contribute to much at the top.  Wrong way to measure.

This slashdot thread caught my attention too (warning, 'strong' language is used).  I do not agree with every sentiment expressed there (and the language wasn't really necessary to make the point) - but the gist I do agree with.  That language/framework is really complex, too complex for many things.

And lastly, a nice peek at XML.



Anonymous Anonymous said....

One thing that I find interesting with Ford is that the Finance people and the Engineering people constantly clashed. Why? Because for Finance to sign off on a new model, it must be cheaper to produce than the previous model. So if you have inflation, you end up reducing the quality and design of your vehicles at the expense of controlling cost. Look at the trouble Ford is in now. Imagine if Mercedes did this? They would look like a luxury Yugo.

Sat Nov 11, 12:30:00 AM EST  

Anonymous Anonymous said....

It doesn't matter what business you're in, if you tie a measurement metric to pay or bonuses, you're going to get what you measure, and probably not a lot more.

Sat Nov 11, 12:27:00 PM EST  

Blogger Alberto Dell'Era said....

The worst mistake is measuring productivity by "mass" of code produced - since it is exactly the opposite, the best programs are the shorter ones almost always.

Being concise is harder then being verbose, in writing and in programming as well.

Sat Nov 11, 01:15:00 PM EST  

Blogger yas said....

I use to track my comments on other blogs. Using its bookmarklet you can easily mark any post you want to track. It even has a Firefox extension that automatically tracks post you commented on. You can even tag your comments.

Sun Nov 12, 06:39:00 AM EST  

Anonymous Anonymous said....

I wonder if the designers of XML actually did any thinking with regards to what could happen if there was any volume of data being transferred.

One would think that it would be a heck of a lot simpler to use a format like [size][tag][data]. I presume, that the parser would not be challenging enough.

Heard of people using XML everywhere. No wonder the system go glacial. The overhead must be tremendous.

Sun Nov 12, 04:01:00 PM EST  

Blogger Thomas Kyte said....

Anonymous right above...

One would think that it would be a heck of a lot simpler to use a format like [size][tag][data].

Ever hear of ANSI X12 or Edifact??

It was even less terse - a file to describe what was coming - very easy, very efficient, decades of proven "workability" over the slowest of dialup lines....

Killed in a month by XML.

Sun Nov 12, 04:06:00 PM EST  

Anonymous Anonymous said....

I agree completely with regards to J2EE frameworks. It is the stuff that happens when people go completely bananas without any real leadership (real life experience?). And the logging stinks to high hell (I have managed a SAP J2EE portal. Not pleasant). Things will go wrong and there is hardly any way of finding out what is wrong.

Those who created J2EE should have known better. There is simply no excuse for messing up so royally. "Death's too good for them" is the quote I want to use to describe what I think of them. They are probably the smartest incompetents on this planet.

Developers know nothing about how to manage anything but their desktops. It usually shows that no system mangers have been involved in straightening out the mess. Or even been involved in avoiding the mess in the first place.

The user interface experience with Java programs is just terrible. It really annoys people where I work.

You know instantly when an application is written in java that you are in for a hard and extremely annoying time.

Sun Nov 12, 04:15:00 PM EST  

Anonymous Anonymous said....

Never heard of those standards. But I used to know COBOL. :-)

The nice thing about the [size][tag][data] format is that it is comparatively easy to handle complex (more or less nested) data structures. It can be used recursively and a decent program can handle items it does not know anything about because it knows the size.

E.g. it could be used for a program to reason about a file path without knowing much about the underlying operating system. Going from logical to physical and vice versa would be something that a library would provide.

e.g. C:\DIR\FILE.TYPE contains a drive element, a directory element and and a name element. Some implementation might encode a file type element as well.

This of course requires a fairly rich supporting library. It might be useful for software which need to work in many environments.


Sun Nov 12, 04:39:00 PM EST  

Anonymous Anonymous said....

On second thoughts. Isn't EDIFACT based on ASN.1? My exposure to that standard is in SNMP. And that was annoying because there was little agreement at the time on how to encode various items. Aaaarghhhh!!!!

Sun Nov 12, 04:52:00 PM EST  

Blogger Noons said....

"That language/framework is really complex, too complex for many things"

I've only been saying that to Oracle, loud and clear, for what, 5 years? And what did that earn me? Ah yes: I'm the "recalcitrant"...

The whole j2ee thing is a shambles, developed by total ignorant incompetents and pushed by people with less experience in IT than a Microsoft security patch!

But it sells middle-ware gear like there is no tomorrow. And it's the most direct way of making a PC with a Cray-class CPU work slower than a 3270 terminal.

So it must be good. No matter what. Did you listen to me? It MUST be good.
PS: see? I didn't even use 4 letter words this time!

Sun Nov 12, 06:37:00 PM EST  

Blogger Gary Myers said....

"Never heard of those standards."
Probably the same could be said of the people who came up with XML. I've worked with EDIFACT. That beautiful concept that, once you've agreed the file structure, you don't need to repeat it for EVERY DATA ITEM.

"see? I didn't even use 4 letter words this time!"
If j2ee isn't a four letter word, you certainly make it sound like one

Sun Nov 12, 07:42:00 PM EST  

Blogger Laurent Schneider said....

It seems with the firefox extension is fine too. I am evaluating both.

Blog comments on seem difficult to monitor with co.mments

Mon Nov 13, 09:06:00 AM EST  

Blogger Joel Garry said....

Imagine if Mercedes did this?

Actually, they did, and they suffered some consequences for it in the last generation S-Class. But it saved them enough money to take over Chrysler, and in the long run their image didn't suffer too badly. Whether Chrysler drags them down financially and they have to do it again remains to be seen.

Nearly all automakers are suffering from massive overcapacity.

21.2 in the XML link is the most important point.

Mon Nov 13, 10:02:00 AM EST  

Anonymous Anonymous said....

everyone involved in software development should read the novel "The Goal". Its about process in a factory. Its remarkably similiar to software development.

Mon Nov 13, 10:17:00 AM EST  

Anonymous Anonymous said....

Reading "On The Psycology Of Military Incompetence" will also help. And "The Inmates Are Running The Asylum".

Mon Nov 13, 11:12:00 AM EST  

Anonymous Anonymous said....

Tom, Java and XML, like COBOL and X.500 before them, are all examples of "design by comittee". They all have problems that stem from a cascading series of compromises designed to "establish a standard". They appeal to the standards people because you can point to a book on a shelf or a document at a web site and say "we have a standard".
PHP, like C, PERL and others, was someone's solution to a problem, not someones need to "establish a standard". They appeal to the implementers among us because they solve problems.
The key, it seems to me, is to find out if the people who developed the language, standard, etc ever solved a problem with it and judge the standard based on the solutions it offers.

Tue Nov 14, 05:06:00 PM EST  

Blogger Scott Swank said....

I like Oracle development. A lot. However, Java is one hell of a slick language. In point of fact, it absolutely blows something like PL/SQL away (except that PL/SQL has this amazing global synchronized memory structure with a Turing-complete functional API -- i.e. the database).

However, Java is plagued with Frameworks. You want to put something on the web, well you have Servlets, JSP, Velocity, Tapestry, JSF, etc (everything but Wicket mostly sucks by the way -- Wicket is so elegant you'll weep). Then there's database access: JDBC, TopLink, Hibernate, JDO, oh my. In short there are so many ways to do things that the libraries overwhelm quickly.

So then fools start hiring people who know the libraries. Those of us in the know refer to these folk as "framework developers." This is quite simply because they typically can't code a lick.

But a good OO developer can really do a nice piece of work with Java. They just have to deal with the "designed by committee" plagues that are most of the J2EE libraries. Except Wicket, it makes the Sweet Baby Jesus (tm) weep.

Thu Nov 16, 02:01:00 AM EST  

Anonymous Anonymous said....


I doubt that committees are that bad. The problem is usually leadership. E.g. Ada is designed by a committee, but there is a single person who decides what goes into the standard. That is why the various versions of Ada are compatible. The various versions does not veer off in strange directions. And contrary to popular belief, Ada is less complicated than C++ (pronounced C, cludge, cludge).

As far as I know the only item they have screwed up on is unsigned types. Later versions of Ada is more or less C compatible. The designer(s) did not know that in C unsigned long is primarily used to get a larger range (e.g. size_t), and not because they wrap. Not being able to map size_t to an Ada type that is not a modular (mod 2^n) is really annoying.

BTW If my memory serves me right, Ada 2005 has interfaces. Largely because they have proven useful in Java.

Sat Nov 18, 08:03:00 AM EST  

Anonymous Anonymous said....

Very good article. This article makes some interesting points. You made a good point, reading your article is a kind of enjoyment. Thank you.
Tactical Flashlights
r c helicopter
video game
Tactical Flashlight
Led Flashlight
rc helicopter
Playstation 2
Playstation 3
Nintendo DS

Mon Jun 23, 03:25:00 AM EDT  


<< Home