Now you are just asking for it.

‘ERD does the one call that causes the hive to regenerate so cleaning up deleted entries and cleaning up the search b-tree. Microsoft mention directly registry fragmentation effecting performance.’

Spare me from your tired gobbledygook already.

From Fundamental of Database System by Elmasri, R. and Navathe, S. B., 5th edn., Addison-Wesley, Boston, pp. 61, 65, 70:

The ER model describes data as entities, relationships, and attributes…

The basic object that the ER model represents is an entity, which is a thing in the real world with an independent existence…

An entity type defines a collection (or set) of entities that have the same attributes…

A relationship type R among n entity types E_1, E_2, …, E_n, defines a set of associations – or a relationship set – among entities from these entity types.

This is the entity-relationship (ER) model. An entity-relationship diagram (ERD) is a pictographic representation of an ER model where entity types is represented with rectangles held together by relationship types represented with diamonds and given attributes represented by ovals [1].

This is a ERD.

This, on the other hand, is not an ERD.

I don’t care if you are a pathological liar or just psychotic – you are wasting my time with your garbage either way.

[1] See p. 80 of the book.

“Nicr try asshole. your still full of crap.”

I’m the only one who will tell you the truth oldman. Actually I don’t really know that. But it’s quite possible that I’m the only one who will tell you to your face, so to speak, what you are really like. And you are right, I am an asshole. Every one is an asshole at some time or another. Some of us are even assholes on a regular basis. But that doesn’t mean we don’t have valid things to offer. And being an asshole does have one important advantage, it makes it brain dead easy to spot another asshole.

I promised Robert that I would curb my use of the word “asshole”, I have regressed here. But the intent of my promise was that I would not call others “asshole” which I believe I have kept (just barely). Also, I did not broach the word in this case. I will endeavor to keep this blog clean. It only takes a little restraint and it’s really only a few of us who stray into the bad language.

So how about it oldman, you want to clean it up a bit?

–nor does an ERD has anything to do with the implementation of a B-tree–
ERD does the one call that causes the hive to regenerate so cleaning up deleted entries and cleaning up the search b-tree. Microsoft mention directly registry fragmentation effecting performance.

The fragmentation that is effecting performance that is not ram cost is the b-tree of the hive not being clean.

That Exploit Guy ERD and other tools like it was a step in the process NT Hive was design and expected to be used with. This would mean the functional requirement to keep the b-tree clean.

ERD is mentioned as solution for windows 2000 to fix the failure of registry hives. It uses an operation on the registry that nothing by default in windows 2000 uses. Interesting enough NT 3.5 has a program using that feature in the default install. It disappears in NT 4.0. Registry problems start from there.

That Exploit Guy this is the problem early NT the registry is operational and safe. Latter generations its not safe by default.

There is a call that causes the library that runs the registry to compact the registry and write a new file clean. That is when it does the b-tree defined concatenation and redistribution. That Exploit Guy. The ERD clean up lists the very function in the library that takes care of the registry that does the concatenation and redistribution.

NT hives has b-tree implemented as a 4 part process not the expected 3 part process.

Microsoft has basically documented the problem and been very vague about it. I suspect they have forgot they made a beast version of b-tree.

‘Go back and provide that ERD link I provided about windows 2000.’

The only link that I remember from you about Windows 2000 is the Wikipedia one that you claim “most people don’t know”. It’s most definitely not an entity-relationship diagram, nor does an ERD has anything to do with the implementation of a B-tree, nor is the Windows Registry based on an entity-relationship model. I am not inclined to bash you with yet anothor textbook, but if that’s what you demand, I am willing to supply.

That Exploit Guy I have already show you the operational example. That leads you to the problem.

This is a shocking part. One operation you watch hives reduce in size as the clean is performed.

–Clean.(this being concatenation and redistribution in the leafs)–

Conatentation and redistributation is performed when you backup the registry hive as per b-tree documentation. Not on delete in on a NT regisrty hive. NT Hives supporting library is design to have the backup hive run every so often with the backup becoming the new registry. So keeping it clean.

The problem here That Exploit Guy is a windows registry hive for the ideas behind all operations is basically a b-tree. Only difference is the delete operation has been split in 2. Its a optimisation. The optimisation is a perfectly valid idea. But if you do that optimisation you have to keep on doing the compaction ever so often. Doing so can result in lower cpu usage under heavy load than a normal b-tree would achieve. Reason if you were smart you would schedule dirty b-tree compaction when the cpu,

Interesting enough its all about IO. Microsoft talks about the registry fragmenting causing performance loss. This is the b-tree filling up with blank space. People forget the main hive is fully loaded into ram. So layout fragmentation in a hive is like SSD fragmentation. Layout fragmentation does not alter hive performance major-ally the lost ram might hurt.

If I delete a leaf in a clean b-tree. I have to record that the leaf location is free to be reused or rewrite the file from then on to fill the space.

By never removing a leaf you don’t have to maintain another table saying where the blank space is. The dirty tree will most of the time lead the insert to blank space. Problem is its only most of the time.

Most b-tree math is theory not real world examples where you have to keep track of memory and reuse.

That Exploit Guy dig out the books of b-tree.

concatenation and redistribution can be done before or after delete. Thing is the NT hive b-tree exploits the fact the wording of many of them does not say how far before or how far after.

Yes a NT Hive is b-tree but not a normal b-tree. It doing something a little warped.

You could say NT Hive does the concatenation and redistribution before delete when the Hive is made. So its concatenation and redistribution with many deletes and inserts performed in middle.

That Exploit Guy the NT Hive format is a very interesting form of b-tree.

If you apply the define of a b-tree lossy NT Hives are b-tree. If you apply the define of b-tree strictly. Then NT Hives are using a new form of tree struct. That is only different by one feature. The fact you can perform many deletes and inserts before concatenation and redistribution is even considered.

oldman
–It is also a FACT that one could get information of all types on that internet. One needed to know where to go, and it was not as extensive IT WAS THERE/–
The key point even with Internet before it was commercial you had to know where to go to find all the other variations of b-tree. If you did not you worked without knowing the other variations existed.

The fact you directly admit it was not extensive. Yes there were a lot of b-tree documentation that never was on the Internet until places starting placing there archives on-line.

Nicr try asshole. your still full of crap.

“More opinion….”

No. You’re wrong. Very wrong. But you are not able to see that you are wrong. You will never be able to see that you are wrong. And that’s not a bad thing oldman. Really, because if you even had a hint of how wrong you are, your brain would melt. Be wrong and be happy that you think you are right. Because in this case, the truth is much worse than the lie.

Nope. “Commercial internet service providers (ISPs) began to emerge in the late 1980s and early 1990s. The ARPANET was decommissioned in 1990. The Internet was commercialized in 1995 when NSFNET was decommissioned, removing the last restrictions on the use of the Internet to carry commercial traffic.[18] The Internet started a rapid expansion to Europe and Australia in the mid to late 1980s[19][20] and to Asia in the late 1980s and early 1990s.[21]“

The Internet was not widely accessible until the 1990s. That’s why Lose 3.1 had no networking and no browser. For the purposes of the discussion, that some algorithm was widely known by a certain name by a certain date because of the Internet, this is very relevant. I am a computer geek and I had no connection to the Internet until about 1992. That’s when I bought my 486. Before that I had an 8-bit processor and DOS as did many others. I did e-mail and ftp by e-mail via bulletin-boards but that is not a mechanism to give universal access to the global community. People were doing lots of stuff in print and it took months/years to publish research. I know. I did that.

Do the maths. How many people were there on the planet in 1997? Billions. M$ provided data that they had 95.7% of the top 80 ISPs signed up to promote IE and the number was 17 million. Look at number of PCs sold. Lots of people had better things to do with their money than own a PC and connect to the Internet. The Internet may have been available to some but it was not universal for many years.

At this point this is no longer totally about the discussion between TEG and the Hamster – it is about a factual error that you are adding to because you want to win an argument.

- It is a FACT that the internet was already well in swing before the advent of HTTP ca. 1991.
- It is also a FACT that one could get information of all types on that internet. One needed to know where to go, and it was not as extensive IT WAS THERE/
- It is also a FACT that ListServs in which all sorts of information including that under discussion was discussed.

My strong suggestion is that you stop making a fool of yourself and let this drop.