The least I usually do is a data model and outline the application using UML. So far he had done by hand or with any vector drawing program or flowchart. The first approach is difficult to maintain because if there are many modifications have to be erasing and repainting the plot being all very sucio.La second approach is cleaner but also very cumbersome.

I have been researching and trying out some applications to improve productivity in this section and although I know there are very good style CASE tool Rational Rose, also I know are very expensive.

I found some tools that are either free or have cost 0, which fits perfectly into my economy.) In particular, I have tried three tools: http://uml.sourceforge.net/index.php Umbrello.

I’ve really been amazed with this tool designed for KDE for its ease of use and careful interface / / even though it does not (or I have not managed to find it) zoom, so working with large diagrams is a problem, which also noticeable when printing diagrams. / /

Updated to the latest version of SID on 03/11/2003, according to the same web Umbrello is not an official version: “this is a (Fairly stable) snapshot of Umbrello from 28.08.2003 and Not an official release Umbrello “and I have checked that already has pleasantly zoom control to charts on screen, but not when printed. This new functionality that Umbrella makes further progress in my preferences.

Another flaw that I’ve found is the lack of freedom in the connections because these are repositioned automatically and your air when moving objects.

On the other hand has good support for code generation from the same program can generate all or part and does it very well.

Debian install is as easy as apt-get install umbrella, but as I said, the package is in SID at the time of this writing, is not an official version, but a snapshot.

Day, rather than a UML modeling tool is a drawing program. Allows flowcharts, network diagrams, etc … particularly through libraries of symbols and a good set of tools. With reference to UML has pretty good support, with a library stocked with the advantage of the freedom of “movements”, the respect he has for the position to have interconnections and control objects on the scale of drawing, make it very manageable.

On the negative side in my opinion is support for code generation, that even if there is through external programs “parse” the file that generates day. There must also say that a good model code be generated only when everything is accurate and complete. But let’s face it, in practice this is rarely the case.

For installation, once again, debianites have it easy, you only have to abuse once more apt: apt-get install and day at the time of writing the article, we will install the version from 0.91 to 10 if our source list point to SID.

Another tool designed to ArgoUML UML modeling is that, although not specific to the Linux platform, is written in Java so it can run on any platform with a Java virtual machine. But it also becomes your worst enemy, because the Java language to be interpreted, the application becomes a very heavy and only usable on very powerful machines, which is not me: (

But despite this setback, is the most comprehensive of the tools I tested, with an excellent and versatile interface to the moment of drawing the model, also enjoying a great generation of code from diagrams.

In closing comment that I opted to use Umbrello by the limitations of my hardware and because, despite its little defects, is that it has been more comfortable in the use and more versatile in terms of code generation.

If you have suggestions or detectais errors in the article do not let the comment. And if you know more of this style applications for use on Linux comentadlo too.

Manufacturers regularly announce new levels of speed and cyber media, unable to understand the technical arcana, are limited to acting as corporate notaries and repeat the texts they refer press agencies. Nobody talks about what can be done with technology, but what technology can do with herself. Welcome to the era of metatecnología.

The other day I found in a story on http://www.wired.com/news/gizmos/0 Red, 1452,47156,00. Html inventing the first autonomous robot capable of operating without human intervention. It is a device that eliminates slugs of crops and, moreover, is capable of surviving with the energy it gets the / / digest / / these invertebrates. During the night, strolling through wheat fields looking for with its electronic eye on their enemies, who holds a mechanical hand and placed in a tank. When morning comes, the robot returns to the barn to make the / / digestion / /, a process that, by chemicals, slugs transformed into energy used at night to continue their work.

Sure to delight the invention many farmers (not the slugs) and the manufacturer may do business with the sale of these devices. It also seems clear that if there were no angry farmers market slugs, the company would not sell a broom. But that does not explain other things: Why not use a device and to locate and blow up landmines scattered across half the planet? A small army of these devices could end quickly with one of the biggest shames of our time. Why not use technology to more than proclaim the supremacy of the chip? “There is no market,” he would say my grandson.

The SOP is based on three principles:

1. Create design standards for the creation and management of exceptions. We got used to that exemptions are a form control structures. Because that is in Java, Ada, C + + and many other languages. But this is not the only possible model. Eiffel is more like exceptions truly anomalous and unpredictable conditions. The basic principle is that a subroutine Eiffel necessarily succeed or fail. The Eiffel exceptions are used to signal that a procedure did not comply with his contract of service. The exceptions are not listed in the signature of the method. The exceptions are represented by strings or integers and there is a class exception.

2. Create tools and / or programming languages that allowed separate control code executing code. For example, selectively hiding pieces of code to read the rest more clearly, and so on.

3. Create standard libraries to generate audit logs and records.

There are thus four major open areas:

1. How to define a policy for creating and handling exceptions.

How many different kinds of exceptions should be created. When should plunge. That part of the program should be responsible for its management.
For example, in Java there is no clear distinction between three types of exceptions: those that indicate a failure (null pointer), which indicates a result (EOF) and to monitor a condition (the input file size too big .) Java Instead it distinguishes between two types of exceptions: those that must necessarily be checked and which not.

2. Closing a standard pre-conditions and post-conditions.

Included within the language and sensitive to whether the program is set to / / debug / / or / / development / /.

3. How to create log files.

Setting the number of messages that are dumped by their severity. How to dump the logs program execution as a multi-threaded web server.

4. How to create useful audit records.

The audit is different from the logs. In the logs are looking for all the details on the implementation of a process. The audit is to determine which user performed that specific high-level operations and when.

Those who are not familiar with the design of complex systems tend to design ad-hoc measures to recover from exceptions. It is the task of good systems architect discriminate different types of abnormal situations and create a treatment policy. It is not the same as a user can not withdraw money because their balance is zero, which have fallen the connection with the database.

One possible tactic is to identify design / / community / risk /. A community of risk is a set of components through which an error can propagate transitively.

For me, another big open question is how close the objects you have open in the case of having to leave in a hurry of a process. The classic case is to release the SQL command and close the connection if an error occurs while traveling through a recordset.

If we look a little further into the future, new techniques of distributed computing (Grid) bring new challenges arising even control an asynchronous errors heterogeneous components, remote and distributed.

If you thought Netscape was evil and oppressive, wait ’till you have a look at Microsoft’s unilateral additions to HTML. They make Netscape look thoughtful and enlightened by their sheer barbarity.

• Use background sounds. Ever so thoughtfully, Microsoft have included the option for the Web author to specify a sound to be played in the background, over and over again. As you can probably imagine, the potential for abuse is tremendous. All you need to do to blast the ears (and possibly the sanity) of your typical MicroSuit is:

  <BGSOUND SRC="SmurfSong.wav" LOOP=infinite>

  La, la, lala, la laaa,  la, lala, la laaa

  ---

• Take advantage of Microsoft’s font extensions. In typical Microsoft fashion, Explorer’s format assumes that everybody of any importance is running Windows, and thus that Windows-specific quirks are standard. Thus they have the FONT FACE option, which not only further desecrates the corpse of content/presentation orthogonality but also assumes that everyone has Arial and Times New Roman.
  The upside of this is that anybody running Explorer will have the Wingdings dingbat typeface. Thus, to monkey-wrench all copies of Explorer, all you need to do is:

  <HTML><HEAD> . . . </HEAD>
  <BODY><FONT FACE="Wingdings">
  
                  . . . your page goes here . . .
  
  </FONT></BODY></HTML>

  And, presto, anybody viewing your page with Explorer or Netscape 3.0 or later under Windows will see nowt but a mass of funny symbols.

  ---

• Use the MARQUEE tag. Who knows what the Microsoft programmers had in mind when they created the MARQUEE tag. A tribute to one of the standard Windows screen savers perhaps? In any case, such a heterodox “feature” presents plenty of opportunity for mayhem. To reduce your Web page to a single scrolling line when viewed with Explorer, do:

  <MARQUEE LOOP=INFINITE>
  
                  . . . your page goes here . . .
  
  </MARQUEE>

Even picking spots at random, the retrieval speed remains unchanged. This is very different from the case with a floppy drive. Bits of information stored next to each other on the floppy can be read quickly, while bits which are recorded a couple inches away on the disk take much longer to retrieve, actual fractions of a second! Sometimes a “D” is added to the front of “RAM,” which makes the word “DRAM.” The “D” stands for dynamic, because the RAM actually forgets what it’s holding if the mother board doesn’t continually refresh (read and re-write) the information. Everything in DRAM disappears when the computer is turned off.

Computers work with very small bits of information, intelligently known as “bits,” which have to be combined together into larger clumps to make any sense. The basic clump used by computers consists of 8 bits and is known as a “byte.” Memory size is measured in bytes, and more is always better. Since computers need huge numbers of bytes to do anything useful, the measure of a “megabyte,” one million bytes, is used. Early 386 machines could limp by on 2 megabytes (MB) of RAM, but 4 MB is the minimum required to run Windows 3.X in a reasonable fashion. 4 MB works fine with most older software, but 8 MB is being required by many new programs. Very few DOS/Windows programs require 16 MB (the next possible increment due to the way memory is installed in most machines), but overall performance does improve with additional memory. Windows 95 can operate with 8MB, but many Windows 95 applications require 16MB, and 32MB is now standard on the run-of-the-mill new PC.

Alarm clocks come in two basic types; digital and analog. An analog alarm clock motor runs directly off the 115 Volt, 60Hz AC power from the wall outlet. Through a series of gears, like an automobile transmission, the alarm clock motor moves the second hand around the face once a minute, the minute hand once an hour, and the hour hand, twice a day. Those of us with a cheap analog alarm clocks and good ears can hear the alarm faintly building up before it goes off. That’s because time, in the analog world, is continuous, so things never happen “all at once.” A digital alarm clock displays the time in Arabic numerals, and changes the time displayed once a minute. If the alarm is set for 6:30 AM, when the clock changes from 6:29 AM to 6:30 AM, the alarm starts to beep. In effect, there has been a step change, an “all at once,” and there are no faint warning beeps before the alarm sounds.

CPUs are digital. The CPU clock speed tells us how many step changes, or operations, the CPU will make in one second. For example, a 386DX-25 CPU does 25 million operations per second. A 486SX- 25 CPU also does 25 million operations per second, while a 486DX/4-100 CPU does 100 million operations per second. CPUs all do millions of operations per second, expressed as “MHz,” so we only need to comprehend the smaller number, 25, 66, 100. The speed of the CPU isn’t the only thing that determines how fast the computer runs your software (imagine putting a race car engine in a truck), but the higher the number, the faster the machine.

There are three other parts to the nomenclature, which you may have noticed above. First is whether we have a 386 or a 486. CPU manufacturers improve their chips with each generation, and the 486 is required by many newer software packages. The 486 replaced the 386, just as the 386 replaced the 286, which was the CPU in the IBM PC-AT. Don’t pay for a 386. Another part to the nomenclature is the “DX” or “SX” part, and the last bit is the “4″ that appears in the 486DX/4-100. With both 386 and 486 CPUs, the DX does complicated math (engineering calculations) better than the SX. Some software may require that you have a DX type CPU, or that you buy another CPU like chip known as a math co-processor to compensate. Just remember DX is better than SX, and don’t buy a machine with an SX CPU unless it comes with a real nice monitor you can use on your next machine. The “4″ that appears in 486DX/4-100 or the “2″ in 486DX/2-66 tells us that while the CPU runs at the stated speed, 100 or 66, it talks with the motherboard it’s plugged into at a lower speed. That means it can work very quickly with information held in the CPU, but if it has to go out and get something from memory, things slow down. Current CPU speeds in new Pentium PCs range from 166MHz (a low end Pentium MMX) to 300MHz for a high end Pentium II.

Before we start to examine things that go in the system box, let’s look at the box itself. The front of the system box is covered by a plastic face-plate, with several rectangular cutouts, a couple buttons and LEDs (lights), a key hole, and a power switch. The power switch is often on the side of the older desktop cases. The rectangular cutouts cover the drive bays, where floppy drives, CD ROM drives and hard drives are screwed in. The larger bays house 5 1/4 ” floppy drives, CD ROM drives, and older hard drives. The small bays house 3 1/2″ floppy drives and newer hard drives. The two push buttons, side by side, are labeled “reset” and “turbo.” The reset button is for re-starting the computer if it stops responding to keyboard and mouse input. The turbo button switches the machine between its top speed, and a slow speed which is only useful for playing games or troubleshooting. On the high speed, the turbo LED should be lit. The power LED should be lit all the time when the computer is on, and the HDD LED would blink when the hard drive is being accessed. The buttons and the LEDs are for the benefit of the user. The computer will work fine even if they are hooked up wrong or broken.

Inside the box is normally where the confusion begins. Here we have RAM, ROM, Megabytes and motherboards, drives and adapters. All of these things are necessary for even the most stripped-down computer, but they need no assembly beyond plugging connectors together and turning in the occasional screw. The cost of each part can be as high as a couple hundred dollars, or as low as $10 or $20. The first part we’ll consider, the power supply, is usually purchased with the system box itself, and in some brand name machines, will have a unique shape (form factor). A unique and painful replacement cost goes hand- in-hand. Power supplies and system boxes are normally sold together, for a price between $40 dollars for a cheap mini-tower to $70 for a desktop and over $100 for a tower. The cheapest cases often have metal burrs inside that scratch unwary hands, and lack fit and finish.

The purpose of this chapter is give you a feel for how all the parts in a PC add up to make a working computer. Like automobiles, many models are built on the same chassis, with the most expensive loaded models costing more than twice as much as the stripped down economy version. There are many “after-market” enhancements you can make, some are “add ons” and some replace the original components. The expensive models have the potential to get you where you’re going faster, and in more comfort. However, there are few fundamental differences between PCs and cars that makes buying and owning a PC a whole lot easier. One difference is that PCs are all built to standards such that their component parts, even across brands, are largely interchangeable. Another difference is that is that most of these parts are either commodities or or available from a single source only, i.e., you can buy memory or a floppy drive from any manufacturer, but Intel chips come from Intel Corporation and MS DOS comes from Microsoft.

There are a ton of different sources for parts to build your own machine. Hit some of the local stores in your area and see what they have in stock and how much they know. You can also pick up a copy of Computer Shopper which has about a million ads in it. Shop around. If you do stuff via mail order, write down everything about the transaction: the date, who you spoke with, whether everything was in stock, what the order number is, who is shipping it and via what method, what the shipping ID number is if available, and anything else you can possibly think of asking. It will help out a lot later down the road if you need to send something back. Don’t order a motherboard from Aberdeen. I ordered one from them, it was defective, sent it back, never got credited, and 4 months later a replacement came in the mail. I’m not exaggerating. If I get around to it I’ll scan the invoice so you can take a look at it yourself. Amazing.

Oh, well since I’m griping here, I might as well add a couple other companies to avoid. Don’t buy some obscure modem because it’s cheap or some Zoom. Spend a penny, it’ll help you out in the end (I run an Internet company, and we sometimes get a customer that has a modem that no one here has ever heard of–it’s a guaranteed piece of junk!). If you’re rich, by a US Robotics Courier, they are the best modems. Oh and never buy anything from Packard Bell.