KCLUG April 2005

53 participants
144 discussions

Re: database
by Jim Herrmann 26 Apr '05

26 Apr '05

Creative Commons baby! Steven Danz wrote: > Wow... you mind if I borrow this? I'll reference you of course. I > used to do the DBA thing long ago, and I've been trying to explain > some of these same things to some DB newbs, and just couldn't get it > to come across this clearly. > > Thanks for your time. > > Steven > > Jim Herrmann wrote: > >> Tim, >> It sounds like you are wanting to do a fairly simple database >> structure, which is a good way to get started. When you say "open >> database format", I assume that you are talking about a target >> platform of a relational database, something like MySQL or Postgres. >> Yes? But that's getting ahead of the design curve. Here's how to >> proceed if you want to follow something akin to a formal database >> design process. I've been doing this for about 18 years as a DBA. I >> hope I'm not volunteering too much information. You said you were a >> newb. :-) >> >> The first thing you need to do is create a logical model. When doing >> a logical model use English business names, and not programming >> abbreviations. Identify all the *Entities* that you need to >> represent, and the *Relationships* between those entities. A entity >> is a singular noun, which would include things like "Person", >> "Department", or in your case, "Reminder". A relationship is a verb >> phrase like "has many", "has one", or "has zero or many". So, you >> can say something like "a Department has many Persons" and "a >> Department has one and only one Manager". >> >> Once you have this, you can create an Entity-Relationship (ER) >> diagram. This gives you a very high level view of your data >> structure. This would be a box for each entity and a line between >> the boxes represents a relationship. Exactly what the box and line >> looks like depends on the modeling tool you use. Then you need to >> identify all the attributes (fields) for each entity. You need to >> identify which attributes uniquely identify an occurrence (row) of >> that entity, and that's your natural key. Keys of parents, of a >> relationship, migrate to the children. If the parent is required for >> a child to exist, then it is an "identifying" relationship, and and >> the parent's keys become part of the child's keys. If it is a >> "non-identifying" relationship, then some or all of the parent key is >> NOT part of the child's key. Now, if you find an attribute occurs >> multiple times in an entity, then you should "normalize" those >> attributes into another entity or entities. There are several levels >> of normalization, and I won't get into it here as it would be a full >> day class, but third normal form is considered the best level of >> normalization for most DB designs. That's where each attribute of >> each entity is dependent upon the key, the whole key, and nothing but >> the key, so help me Codd. (E.F. Codd, an IBM fellow, created >> relational database theory.) :-) >> >> Once you have a fully attributed ER diagram, you are ready to create >> a physical model based on the logical model you created. This is >> where you create names that you want to use in your program, as you >> turn entities into *tables* and attributes become *columns* and >> relationships become *foreign keys*. You might shorten "Department" >> to "DEPT", for example. It's best if you use standard abbreviations, >> meaning be consistent, as much as possible. It makes programming >> much easier. The length of table and column names may need to be >> shortened, depending on the target database. That's part of the >> physical design process. You also need to decide what indexes will >> be on your tables. Some for uniqueness, such as on the key, and some >> for performance, depending on how you are going to access your data. >> >> There are also performance and convenience tricks that you can do >> when you take your model physical. For example, if you have some >> natural keys that have lots of columns, you might introduce >> artificial keys, which would be something like an integer or reverse >> timestamp. You may also at this point decide to "denormalise" some >> entities to make your system go faster, but that comes at a price of >> maintaining redundant data, and it breaks Codd's rule. Depending on >> the DB platform, there are lots of physical implementation issues to >> consider, if you are building something with lots of data or lots of >> transactions, but you probably don't need to concern yourself with >> all that for this project. >> >> Hopefully you find this information helpful. If you have any further >> questions about data base, I'd be glad to help. Sorry about the >> brain dump. ;-) >> >> Enjoy, >> Jim >> >> Tim reid wrote: >> >>> I'm thinking of programing/writing/whatever a database to keep track >>> of some information for work. I think that I would like to keep track >>> of 7-12 fields of information, and be able to set up reminders that >>> are tagged to date information. For instance, "you did such and such >>> 30 days ago...you can now file the claim, here is the information that >>> you recorded earlier." >>> >>> How would i go about doing this, free as in beer, in an open database >>> format? Let me also mention that I'm a total newb ;) >>> >>> TIA, >>> >>> Tim >>> >>> >>> >> > > > -- Progressive Values ARE American Values Responsibility, Empathy, Freedom, Opportunity, Prosperity, Fairness, Trust, Honesty, Open Communication, Community, Cooperation http://musicalprogress.org/

1 0

Vim/vi scroll-margin
by Gerald Combs 25 Apr '05

25 Apr '05

Is it possible to force vim/vi to not place the cursor at the bottom of the screen when searching? For example, if I search for "some_func()" and match the example below, vim will more often than not place the cursor and the first line of some_func() at the bottom of the screen. If I want to see all of the arguments to some_func, I have to scroll down. Pine has a setting called "scroll-margin" that forces the cursor into a zone <n> lines from the top and bottom of the screen. Does vim have something similar? some_func(a, really, long, list, of, arguments, that, spills, over, into, several, lines, and, will, likely, be, hidden, from, view, thereby, requiring, a, bit, of, scrolling);

2 2

Re: UML CASE tools
by Jack 22 Apr '05

22 Apr '05

--- Jim Herrmann <kclug(a)itdepends.com> wrote: > Our company uses Rational Rose for UML modeling, and > ERwin for DB > modeling. But now you're talking serious money for > either of those > tools. Probably not what you're looking for, but I > thought I would > offer it up. Have you looked at Dia? It will draw > boxes that you can > connect. You can't generate anything out of it, but > it's got the > boxes. Thanks Everyone for the responses. I went with Dia + tedia2sql for the time being. With Dia you can use UML class objects to do the drawing (turn off the display of functions. Adding attributes for the fields. You can set Primary keys by setting them as protected elements. Tedia2sql can read the dia document and create sql scripts that can be used to create the database objects with primary keys and foreign keys if you want. It can create for many db formats and can handle quite complex set ups. Not quite the one step type of operation I was looking for but a real big time-saver. Brian D. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

2 1

Re: Screen Resolution Trends
by Jared 22 Apr '05

22 Apr '05

> But then, I'm still waiting for the browser that renders vector graphics. > Regards, > -Don Don: All browsers already do. http://www.walterzorn.com/jsgraphics/jsgraphics_e.htm It's a matter of time before someone couples this concept with SAJAX and creates streaming vector graphics in an ordinary (no plugins) browser... using simply JavaScript. Video games in your browser will go far beyond this level: http://www.def-logic.com/games.html Get ready. -Jared p.s. sorry about the IE warning--IE's rendering engine is faster than Gecko (for this kind of work) at the present time, so they targeted it.

3 2

RE: Linux Certification
by Jack 22 Apr '05

22 Apr '05

--- Leo Mauler wrote: > Don't know much about this being a good way to find > out about prospective employees. Some of us are > very > good at bluffing: about four years ago I had never > touched Microsoft Excel, and yet I scored 88% on the > test for it for some temp agency. > If you scored 88% on a programming test, I certainly wouldn't hire you. That would mean that someone would have to fix more than 10% of your code. Not an exciting prospect. But I was actually referring to tests that I would create using test building software that comes free with Linux. I think tests should be created specifically for the needs of the company doing the hiring. I haven't been to any of the "prove it" test sites. Certainly, they are no magic bullet. But a certificate only says you studied some book, or took a course and then took a test and can be relied upon to perform any task identical to the test. Of course the performance tests are better than this. Multiple choice tests are the worst measure. Brian D. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

1 0

Re: Screen Resolution Trends
by Leo Mauler 22 Apr '05

22 Apr '05

This is certainly true of browsing the web, I've found very few websites which don't demand a maximized window to view the page properly. I suspect this is the natural consequence of ad-supported websites: the ad-supported sites shove content out of the way to display ads in less than a maximized window environment; the majority of websites are ad-supported; so naturally web designers design their pages to the lowest common denominator, which demands a maximized window. --- Jonathan Hutchins <hutchins(a)tarcanfel.org> wrote: > Something that's been skirted in this discussion is > the idea that the whole point of a windowing > environment is that no one application takes all > of the resources - they share the screen, so > whatever the screen's maximum resolution is, any > given app should take _less_ than that for it's > standard resolution. > > That's been something that's frustrated me. We > have this ideal of multiple working windows, but > unless you're on some super-pricey graphics > workstation you're usually reduced to maximizing > each application in turn to make the most of your > resources. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

1 0

Re: Linux Don'ts
by kulua＠linux2themax.com 21 Apr '05

21 Apr '05

Justin Dugger <jldugger(a)gmail.com> wrote .. > I don't know about SCSI, and I've only heard about ATi cards 3rd hand > (from a lot of people). The big problem with telling your nephew what > brands to avoid is that the PC hardware business has multiple tiers in > most cases. One company designs the chips, another one makes them, a > final one puts them on a board and markets them. Linux drivers > generally support chips, not brands. For example, my wireless card > uses the Atheros chipset, and the madwifi driver supports most all > atheros based cards. Thing is, you can't always look on the box and > figure out the chipset. > Linux has enterprise support I would say that most scsi especaly older already has support. some may have been droped due to sheer age of equipment but should pretty much all be there. Ati=VERY BAD. they refuse to support their products they are giving a token effert on the display drivers and refuse to support the video capture at all. They have drug their feet at getting info to the gatos project. The mini-itx is also a intresting area. hit and miss support for devices and video but pretty good allaround. jason whats wrong with nforce nicks? other then being a pita to compile drivers. Intel nicks are the way to go in my opinion for any thing less then 1000tx but other then minor squables with hardware most is supported. you can also tell him to look at the forums of the manufacturers ie for system boards and see what peopel are saying there about linux and that componant. I have a dwlag 520 & 530 atheros. I havent messed with the 530 in linux yet due to a issue with 530's hal. I was wondering if you noticed a signal strength difference between windows and linux. Im not getting strong signals from my 520. my desktop can see the linksys router but not the linux server. I coud also be configuring it wrong.

1 0

Red Hat Clustering
by Dave Hull 21 Apr '05

21 Apr '05

Anyone on the list using Red Hat EL in a high availability cluster environment? -- Dave Hull http://insipid.com

2 2

Screen Resolution Trends
by Don Erickson 21 Apr '05

21 Apr '05

Here's a bit of data that web users/designers may find interesting. This is a table of (rounded) screen resolutions and months that the data was collected. I've had a javascript function to tailor pages to the screen resolution of visitors to a few web sites, and "databased" the results. Here's how it breaks down: 2004-03 2004-10 2005-01 2005-03 640x480 2 % 0 % 1 % 1 % 800x600 34 % 30 % 24 % 23 % 1024x768 54 % 55 % 57 % 58 % 1200x960^ 11 % 15 % 18 % 19 % So, by these numbers the ubiquitous 800x600 screen resolution that pixel-based programs have to design for are now used by fewer than one quarter of the users out there. I think that this is significant. Also, on the certification topic, is LPI still an entity at all? For distribution-neutral certifications, I thought that it used to be the pinnacle, or at least they intended to become that, but it didn't even merit a mention in a dozen or so emails on this list. I do agree that a task-based test is going to tend to be a more valid measure of skills than a multiple-choice, but I thought the LPIC-2 was pretty tough. Regards, -Don

5 7

RE: Linux Certification
by Leo Mauler 21 Apr '05

21 Apr '05

I smell "Catch-22": certs are useless, don't get them first; but there's no other way to prove to an employer that you know something...other than getting a job which requires that you prove what you know, except you can't without a cert which is useless to get the job you need to prove you can do something... I mean, the only alternative is to occasionally take two years off every decade and re-take an A.A.S. in IT (guess what I'm doing...). Personally, I'd think that any "hands-on" cert is still worth something. This would include Cisco and RHCE/RHCT. There at least you can point to a cert and say "I had to demonstrate real-world knowledge to get it, not just write down answers on a test." Granted, Cisco is probably an employer-sends-you-to-it cert, at about $1500 to take the whole thing, but you can still get that RHCE/RHCT cert for less than $300. --- Jonathan Hutchins <hutchins(a)tarcanfel.org> wrote: > I've been watching the certification saga, both > Linux and otherwise, for a while now. > Certification is something you do within a position, > not as a qualification for a new position, not as a > ticket to the next job. > > I think that's where it remains, and when someone > asks "what's a good certification to go for?", I > say "Whatever your employer will send you to". > If you don't have the qualifications to get the > job without the cert, if the prospective employer > isn't convinced that certification is a formality, > for you, and isn't willing to pay for it, I think > it's a bad gamble. > Get the job first, then worry about the > certification. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

2 1

← Newer
1
2
3
4
5
6
7
8
...
15
Older →

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

KCLUG April 2005