Desktop computer

[Skuzzy]

There is more than a good chance the software raid is marking the cache dirty on each access.  This is a cheap way to insure drive integrity in a raid.

None of the standard Windows library functions, which read data from a device, can avoid going through the system cache.  Then again, 8GB of system RAM may not be enough RAM to contain the cache if the game is taking to much memory.  There is some speculation, on my part, as I am not familiar with the game, nor the system being used.

I do know my home computer is loading all data from system memory, once a game has loaded it before.  Easy test using Aces High.  Before going "Offline", pick a big terrain, such as ozkansas.  Go "Offline".  It should take a bit.  Exit, then go back in again and it will be faster, if your system is really making use of the Windows cache.

[MrRiplEy[H]]

There is more than a good chance the software raid is marking the cache dirty on each access.  This is a cheap way to insure drive integrity in a raid.

None of the standard Windows library functions, which read data from a device, can avoid going through the system cache.  Then again, 8GB of system RAM may not be enough RAM to contain the cache if the game is taking to much memory.  There is some speculation, on my part, as I am not familiar with the game, nor the system being used.

I do know my home computer is loading all data from system memory, once a game has loaded it before.  Easy test using Aces High.  Before going "Offline", pick a big terrain, such as ozkansas.  Go "Offline".  It should take a bit.  Exit, then go back in again and it will be faster, if your system is really making use of the Windows cache.

It's win7 native software raid, don't know if it invalidates the cache or not. One would think win7 can manage itself.

[Skuzzy]

You know what they say about assumptions.  It is, after all, Microsoft.  It is pretty easy to test.

[MrRiplEy[H]]

You know what they say about assumptions.  It is, after all, Microsoft.  It is pretty easy to test.

Yep but we'll have to wait untill my kid stops playing on the thing!

[Brooke]

I guess you've been pretty lucky since XP is suspectible to rapidly spreading worms especially when combined with java and internet access. Are those more like embedded systems where users practically never actually access the desktops? If you need uptime, how do you perform updates? Most of them require a reboot.

I would be crapping my pants if I was responsible for a system like that lol. Best of luck!

I don't think it's just a fluke that it works well since we're talking about over 100 machine-years of data here.  The Java isn't for the browser -- it's the VM for running the application (consider app + Java VM to be the same as an .exe).  Users are not usually going to be using the browser on these systems for much, and they aren't running Outlook or MS Office or much of anything other than the custom app for running the equipment.  We need uptime when they are running the process, but there are times between synthesis runs when maintenance and reboots can happen.  However, since most installations are at places without much IT support, they just get left on and rebooted if the machine is telling an operator that it needs to be rebooted.

If this setup were susceptible to much and had resulting problems, we (as the manufacturer and seller) would get complaints.  I don't remember a failed run ever being a result of OS or software crashing at any customer site.  Again, that's not to say that it never happened at any site around the world in the past 5 years, just that it is so rare that I don't remember any.

Alternatives to Windows XP plus Java as the application language all would have cost far more (like double or more) in development resources to get the same level of functionality and uptime; or to surpass the uptime of this system (which again is historically nearly perfect) would have cost several times the development resources we used for very little extra benefit.  It's a much different environment of usage than a web app server, db server, etc.

But even for our desktops for workers, those are left on all the time, rebooted only when needed as part of regular updates or as a first step of fixing some software quirkiness (which isn't that often).  My experience there is in the 500+ machine years of operation without much trouble in regard to leaving them on.  Yes, there is the clueless, bumbling contingent that manage to screw up their computers by installing crapware and the occasional problem because of Outlook (which generates the most trouble of any application we had in common use), but that's not the fault of the OS.  And I leave on my home system for weeks at a time before it is rebooted without any trouble.

[gyrene81]

i just totally despise java apps haven't seen one yet that isn't a clusterf@*k that makes my system run like toejame.

[Brooke]

i just totally despise java apps haven't seen one yet that isn't a clusterf@*k that makes my system run like toejame.

Are you talking about Java aps or Java applets?

Either way, you can use any language to create crapware -- that is not an aspect of the language.  There is plenty of crapware written in C, for example, yet C is not a bad language.

[gyrene81]

poorly written c programs still run better than anything written in java...at least everything that has been publicly available to see in the past 25 years.

[Brooke]

poorly written c programs still run better than anything written in java...at least everything that has been publicly available to see in the past 25 years.

I suspect that neither of us has tried every program publicly available in the last 25 years.    Also, Java apps weren't around much prior to about 2000, and the first JDK wasn't even available until 1996.

I have used vastly more apps written in C than in Java (probably 100's of times more) and hence have seen many more horrible ones in C than in Java.  Most of my experience with Java apps are IDE's written in Java (which were excellent) and bioinformatics tools written in Java (which were excellent).

Again, whether or not an application is horrible depends foremost on the person's programming, not the language.  Granted, some languages are less prone to programmer error than others.  For example, writing in assembly is more prone to programmer error than C, and C is actually more prone to programmer errors than Java (as Java is more strongly typed, isn't based as much on raw manipulation of pointers, etc.).

Pointers in C historically were a source of a lot of inadvertent programming errors.  Back in the days of DOS, this was a huge pain in the neck, since DOS did not have protected memory.  So, you could have an errant pointer stomp all over the place -- you might stomp on the code for the printf statement, for example, and not know it until your program crashed on a printf statement, which is nowhere close to where the bug in your code is.  That very event back in the early or mid 90's (writing neural network code for financial analysis) was what caused me to make a holy vow never to write another DOS program again and made me switch to OS/2.  Windows 95 and Windows NT weren't out yet, and my choice was realistically DOS, OS/2, or Unix, which at the time was very expensive and hugely cumbersome on the PC.  I guess one option was the Mac (which I programmed in during grad school), but it was not very friendly to scientific programming in those days for various reasons (although I picked Macs for our office staff, on which we played Marathon after hours -- man, that was fun).

Every language has a set of things it is good at and some things its not good at, too.  C has the widest set of uses.  Java is great at some things, too.

[MrRiplEy[H]]

I suspect that neither of us has tried every program publicly available in the last 25 years.    Also, Java apps weren't around much prior to about 2000, and the first JDK wasn't even available until 1996.

I have used vastly more apps written in C than in Java (probably 100's of times more) and hence have seen many more horrible ones in C than in Java.  Most of my experience with Java apps are IDE's written in Java (which were excellent) and bioinformatics tools written in Java (which were excellent).

Again, whether or not an application is horrible depends foremost on the person's programming, not the language.  Granted, some languages are less prone to programmer error than others.  For example, writing in assembly is more prone to programmer error than C, and C is actually more prone to programmer errors than Java (as Java is more strongly typed, isn't based as much on raw manipulation of pointers, etc.).

Pointers in C historically were a source of a lot of inadvertent programming errors.  Back in the days of DOS, this was a huge pain in the neck, since DOS did not have protected memory.  So, you could have an errant pointer stomp all over the place -- you might stomp on the code for the printf statement, for example, and not know it until your program crashed on a printf statement, which is nowhere close to where the bug in your code is.  That very event back in the early or mid 90's (writing neural network code for financial analysis) was what caused me to make a holy vow never to write another DOS program again and made me switch to OS/2.  Windows 95 and Windows NT weren't out yet, and my choice was realistically DOS, OS/2, or Unix, which at the time was very expensive and hugely cumbersome on the PC.  I guess one option was the Mac (which I programmed in during grad school), but it was not very friendly to scientific programming in those days for various reasons (although I picked Macs for our office staff, on which we played Marathon after hours -- man, that was fun).

Every language has a set of things it is good at and some things its not good at, too.  C has the widest set of uses.  Java is great at some things, too.

I noticed that you talk about unix but disregard linux completely. It would have been a very easy and no cost solution for example 5 years ago, even for running the java app.

[Brooke]

I noticed that you talk about unix but disregard linux completely. It would have been a very easy and no cost solution for example 5 years ago, even for running the java app.

Our choices for this were made closer to 2000, but regardless I wouldn't choose Linux for this particular application.  That's not because I think it's bad -- it's good for many things.  In fact, we ran an important part of our company on Debian.  It has more to do with the ability for customers to easily get local third party support (to deal with computer problems if needed), drivers for various cards, and ability to run other software packages.

My other mention of Unix was with regard to PC's that ran financial number-crunching programs for another business I was in.  That was in 1993 or 1994 when I was choosing.  I ended up choosing OS/2, as the non-free Unixes out there (like Xenix) were expensive and very cumbersome to deal with, and the free Unixes didn't seem polished enough yet to me (not that I was the greatest expert on every Unix available, though).

I was very happy with OS/2 and wished it to win out over Windows NT and Windows 95, as at the time of their introductions, to me OS/2 seemed very much superior to either one.

[MrRiplEy[H]]

Our choices for this were made closer to 2000, but regardless I wouldn't choose Linux for this particular application.  That's not because I think it's bad -- it's good for many things.  In fact, we ran an important part of our company on Debian.  It has more to do with the ability for customers to easily get local third party support (to deal with computer problems if needed), drivers for various cards, and ability to run other software packages.

My other mention of Unix was with regard to PC's that ran financial number-crunching programs for another business I was in.  That was in 1993 or 1994 when I was choosing.  I ended up choosing OS/2, as the non-free Unixes out there (like Xenix) were expensive and very cumbersome to deal with, and the free Unixes didn't seem polished enough yet to me (not that I was the greatest expert on every Unix available, though).

I was very happy with OS/2 and wished it to win out over Windows NT and Windows 95, as at the time of their introductions, to me OS/2 seemed very much superior to either one.

On 2000 most of the linux distros were still for hardcore users only

I bought my first linux cd in 1998, Caldera linux. I didn't even have a broadband connection at that time. It wasn't as easy as the distros now if you know what I mean

[save]

Started with slackware 1.2 , think it was 3 diskettes 

[gyrene81]

I suspect that neither of us has tried every program publicly available in the last 25 years.    Also, Java apps weren't around much prior to about 2000, and the first JDK wasn't even available until 1996.

ok then more like 16 years...    i'm 52, it's easy to forget a few years when there are so many to remember these days...   

I have used vastly more apps written in C than in Java (probably 100's of times more) and hence have seen many more horrible ones in C than in Java.  Most of my experience with Java apps are IDE's written in Java (which were excellent) and bioinformatics tools written in Java (which were excellent).

i know what you mean. there are some niche areas where java (well written) is very useful but, where we run into it on a regular basis, someone just tossed some poorly written crap out there and it causes problems. when i started all this silly fascination with computers, it was c and cobol that forever broke me of the desire to pursue a career in programming. probably would have been a different story if i had gone to college and learned it in an academic environment instead of on the job. where i'm at now sooner or later i'm going to have to buckle down and at least learn perl and python...i don't have the patience to try and memorize the million ways to do things.

[Brooke]

probably would have been a different story if i had gone to college and learned it in an academic environment instead of on the job. where i'm at now sooner or later i'm going to have to buckle down and at least learn perl and python...i don't have the patience to try and memorize the million ways to do things.

I don't think college has a large advantage to learning to program or being good at it.  I've built software-development teams, hired software developers and IT people, and then worked with them (i.e., my hiring was as part of the working team, not as an HR person who has no idea if his hiring really worked well).  Of the best developers, some majored in computer science in college, some majored in something else in college, and some didn't go to college.  I think that learning on your own or on the job can be a great way to go, too.