Blog All Title

You've always hear about the five stages of grief, and there is no one currently experiencing more grief than the Republican Party. With the low Presidential approval rating, the fiasco in Iraq, Woodward's new book, and now with the Foley affair, they have plenty of grief to go around, and it's almost fun to see them go through the various stages.

First, we had Bargaining as KIrk Fordham tried to bribe ABC News into not publishing the IMs from Foley.

Then, we had Anger as various Republicans started blaming each other for the mess.

Now, we are getting into denial. According to James Dobson and Matt Drudge, the Foley affair NEVER happened. It was a joke. Yeah, that's the ticket. A prank. One of the pages must have made Foley do it!

However, some Republicans have already come to the final phase, Acceptance. According to this Robert Novak article, an unnamed Republican says that Representative Hastert's resignation from Majority Leader in the House is really moot.

We are sure to lose the house, and [Republican Majority Leader] Denny [Hastert] never would want to be minority leader.

Somewhere depression is suppose to come in, but I suspect that will hit the Republicans after this November's election.

Have you seen those super automatic espresso makers? These machines things cost as much as $3,500. Who is going to buy one of those?

I would for one. In fact, I've bought two. A few years ago when I was a theoretical Internet millionaire, I bought my mom one for her birthday. I bought a second one for myself not too long ago.

These machines are amazing marvels of technology. Coffee beans go in one end, and after the machine makes a few rather terrifying noises, out comes espresso on the other end. The machine automatically grinds the beans, tamps them down, and brews the coffee. A cup of espresso comes out one end, and a little coffee doot is ejected and goes into the coffee doot holder that has to be emptied periodically.

A good coffee maker will make a cup of espresso with a good rich creme (Italian for coffee scum). The funny thing is I actually didn't drink coffee until about eight months ago.

My son Daniel was the first coffee drinker at our house. He was getting up at 5:30am to get to his high school and didn't get home until five or so at night. He started drinking instant coffee, and my wife was concerned with it. So was I. I told him right out:

Daniel. I don't want to see a young teen sitting there drinking instant coffee. You might as well learn to drink the good stuff. I bought him a drip coffee maker and later on a coffee grinder, so he didn't have to drink Folger's which according to the Geneva convention, cannot be served to prisoners of war. I soon got a job which made my son's hours seem quite reasonable, and tea was just not cutting it.

I quickly learned the wonders of coffee. I use to drink coffee, but only with sugar. When I was diagnosed with diabetes, I decided to give up all sweetened beverages. I quickly got use to unsweetened tea, but never got use to the taste of unsweetened coffee. Now, a half dozen years later, and I decided it wasn't that bad after all.

The secret is decent coffee. The stuff at my job is awful, and I won't drink it straight. It's thick and bitter with a bad burned flavor. I found that if I started with decent beans and ground them myself, I could get a decent cup of drip coffee.

My downfall was a restaurant near my job called Cafe K. I had ordered a cup of Latte, and loved it. I have ordered latte before at Dunkin Donuts and it was pretty much coffee with a lot of milk and some bubbles on top. This however, was wonderful. It was rich and smooth with a deep coffee flavor. The texture was creamy and a consistent finely textured foam throughout. I ordered espresso and it was strong without being bitter or burned with a red creme resting on top. This was no coffee as much as Java based heroin. I was completely hooked.

Right before Passover, I bought a refurbished Saeco automated coffee maker from Whole Latte Love and I'll go through a couple of cups of coffee per day, and I'm not even around during the day. We are thinking of getting Costco club membership just so we can buy beans in bulk. Even my youngest uses the machine, although not for making coffee. He uses the steamer to steam milk for hot chocolate.

I am still working on my milk steaming skills. I have a pretty fine foam texture on my milk, but I am missing something. At Cafe K, the coffee and milk mix into a luxurious foam. However, I can't seem to get the texture right. Mine seems to separate. It looks okay, but its not the same that I can get at Cafe K. I am thinking of quitting my job and working at Cafe K just so I can learn how to make a decent latte.

For over 14 years, I've been a ClearCase administrator, and ClearCase has been very good to me. However, lately I've been telling all of my friends that ClearCase is quickly becoming a dead end version control system, and that they should start learning Perforce. I believe that Perforce will become the most important version control system, and it is all thanks to Subversion. To understand why ClearCase is in trouble, you have to understand Cado Computers.

Cado Systems built business computers way back in the late 1970s and early 1980s. The basic Cado 20/24 system was based upon the eight bit Intel 8085A and came with 48K (that's Kilobytes, not Megabytes) of memory. Despite this, Cado Computers could handle up to four users simultaneously. While similar computers like the IBM Datamaster were over $100,000, a typical Cado system could be had for around $20,000. That was an amazing price breakthrough. A typical small business could finally afford a real computer! They could easily save the cost of a Cado Computer in just a single year by having the computer handle much of the accounting and ledger keeping.

A Cado Computer can track invoices, and help keep money flowing into your organization. It could print your invoices, checks, and bills automatically. The closest competitors couldn't produce a computer with this feature set for five times the price! So, how was Cado able to produce a computer that could do so much for so little money? Simple, Cado realized that hardware was much more expensive than software. By optimizing the hardware, Cado pushed everything off to the people who handled the software. For example, Cado had no file system. If the developer wasn't careful, they could easily place two files over the same area of the disk, or put a program on top of a file. The developer actually wrote down on a piece of paper where each file and program was stored on the disk.

Cado saved hardware costs by allocating only 2K for each user for program space. In order to fit programs into that tight amount of memory, developers were given few resources in order to program. Developers only had 26 six byte numeric registers, nine alpha registers that ranged in size from 20 bytes to 40 bytes, and five 255 byte long buffers for data storage. All these resource were shared by the developer and the OS, so the OS might decide to use a particular buffer which meant that any data the developer saved there would be overwritten.

Sure development was slow and difficult, but as long as hardware costs were high, it made sense to push these tasks onto the developer. In fact, the whole architecture of a Cado Computer was built upon the premise the Hardware was always going to be more expensive than software. You could say that Cado Computers bet the company on this fact.

By the mid 1980s, that assumption broke down, and Cado quickly saw its sales plunge. PCs were dirt cheap and plentiful. Software tools like Dbase made developing complex business software much easier. When PCs prices really dropped, and millions of PCs were out on the market, PC accounting software could be written that was more flexible and more powerful than could fit on the anemic Cado hardware. The cost of developing that software could be spread over millions of PCs. By the late 1980s, you could by a PC based accounting system for under $10,000 that could outperform anything found on a $20,000 Cado. Cado simply disappeared in a wave of sales and mergers.

ClearCase got its start almost 20 years ago by a little company called Atria. Most version control systems are fairly similar: The user creates a working directory on their local computer, and then retrieves the version of the software they want from the version control system. The user has to keep retrieving the software, and has to make sure they have enough disk space to get all of the files they need. Compiling the files into a software package was a long process since most computers had slow processors.

ClearCase changed all of that. Unlike most version control systems, ClearCase used views that could be stored on a large and very fast server instead of a slow and small desktop system. To a typical ClearCase developer, once you set up your view, it appeared that you directly accessed the software from the source archive. The slow process of copying all of the files to your local workspace was no more. Better yet, you really didn't need a local workspace since your view and the smoke and mirrors of the VOB handled what versions of the software you would see.

And, ClearCase also had another magic trick. Why bother compiling code if someone else already compiled it for you? What if the version control system looked at what you wanted to build, said to itself, "Hey Ralph has already built this piece of code!, why not simply give you what Ralph built instead of you wasting your time compiling it yourself?". This was called winking in and was one of ClearCase's most highly touted features.

In the past 20 years, things have really changed. That slow and small desktop system contains a really fast and powerful processor (maybe even two), and now you've got at least 40 or more gigabytes of space sitting on your desktop. Meanwhile, networks have become more ubiquitous and more crowded. Now, ClearCase isn't the only thing that wants to use your network, and the heavy traffic used by ClearCase bogs everything down.

To handle the change in environments, ClearCase introduced what are known as static views. Static views download all needed files to your local system where you can do your development. Of course, this simply makes ClearCase just like any other version control system, except that ClearCase costs more, takes more resources, and is slower. Besides using snapshot views means you can no longer do winking in already built objects.

Except that ClearCase's winkin feature isn't as important as it once was. You see, that powerful piece of iron on your desktop compiles code in no time flat. By the time ClearCase searches its database, finds a view that has the same code you're compiling, and then transfers that to your view, you've already built that code. What takes long now is linking all those little pieces of code together, and that takes quite a while thanks to the much more complex structure of the software and underlying API. And, linking over a network is much slower than linking on a local machine.

So, you see that like Cado, ClearCase made basic assumptions about its environment that are no longer true. However, Rational and later IBM have found a way to keep ClearCase sales up. Developers may hate ClearCase, but IT departments love it.

In a large corporation IT departments want to control development, and ClearCase's centralized structure makes it easy to do. ClearCase is very complex to setup and use and local expertise is expensive. However, if you have a centralized IT organization, that cost of expertise can be shared among all development organizations. Rational Software (and later IBM when it bought Rational) understood this, and was able to encourage ClearCase sales by making ClearCase part of a larger IT strategy. Fortunately for Perforce, Subversion was released.

Subversion was released about three years ago as an open source version control system. It was pretty much everything ClearCase was not. Where ClearCase was extremely expensive, Subversion was free. And, where ClearCase was hard to setup and understand, Subversion was easy to use. Although Subversion was not as feature rich than ClearCase, it was much easier to use, and being free meant that anyone could down load it and use it. And, that's just what happened.

As I said, developers hated ClearCase as much as IT loved it. And, what developers really hated was the way IT could control the use and workflow of ClearCase. Want to start a new project? You'd had to contact IT. Add a new user to your development group? Contact IT. Change something around in your development? Create a new branch? Add new files? IT needed to approve it. And, IT's approval could take days. However, what if the development organization could bypass IT by downloading their own version control system. A system that is freely available, and easy to setup and install? Thousands of development departments broke their corporate policy, bypassed IT, and setup their own Subversion development environment.

When IT found out, their typical reaction was try to stomp it out. Unfortunately, you couldn't. How can you prevent someone from downloading and using a freely available piece of software? And, when push came to shove, local managers would side with their development departments. After all, their department needed the software, and bypassing IT allowed more flexible development and less downtime.

In the last year or two, IT is recognizing that you cannot force developers to use ClearCase anymore. However, there are legitimate concerns with local development shops using Subversion. First of all, most of these Subversion archives are sitting on people's desktop machines and not getting backed up. Not something you want for important commercial assets. Second of all, Subversion really was missing some important features. For example, Subversion does not really have built in account protection. It depends upon the operating system or Apache server to provide it. Subversion also didn't allow users to delete individual versions of a file (or entire versions of a file). Normally, this was a good thing, but what if a user checked something into an archive that shouldn't be there such as proprietary information?

There were also problems scaling Subversion, and with Subversion's ability to handle more complex development strategies. Plus, who is IT going to call if they have a problem with Subversion? This is where Perforce comes into play. Perforce is fairly straight forward to use and liked by most developers. Perforce is also more feature rich than Subversion. Plus, unlike Subversion, there's a company you can call for support. Although Perforce is more expensive than Subversion, Perforce is much, much cheaper than ClearCase.

The results is what I call the great compromise. IT gives up their ability to control every single aspect of CM. In return, local developers allow IT to take care of the iron, training, and other tasks that local development shops never liked in the first place. Developers have a fast and efficient system for development, and IT has secured the corporate software assets from disasters. All in all, not really a bad compromise.