MySQL's the fast but clumsy child, whereas Postgres is the slow but methodical child. MySQL tacks on features and calls them stable yet you can't trust your data to it. Postgres is ver conservative out of the box but each release includes significant speed increases.
MySQL is very fast for super-simple selects and inserts. Throw anything more at it and it crawls. Postgres, on the other hand, is pretty conservative in terms of memory usage but once you increase the shared memory allocation and turn off the forced-fsync it handles much faster than MySQL for even slightly complex queries and will outstroke MySQL handily as you pile on connections and complex queries.
MySQL's not had the features because a) they slow the DB down and b) their entire philosophy is "we just store data. You make sure it's good before you give it to us or we'll make it good in our own way and not tell you."
MySQL 5 is a HUGE leap forward for MySQL and most of the points will probably become moot. Let's just hope they fix the default-value fiasco also...
Are they planning on fixing the artistic license that MySQL routinely takes with the data thrown at it? I'm talking about autotruncation, auto "converting" strings to numbers, NULL and 0 being the same thing... Until MySQL takes data consistency or at least validation seriously it will never make it in this shop.
When MySQL 5 is released PostgreSQL will get some more Open Source competition and that is a good thing.
Totally agreed. I'm a fan of Postgres, this is true, but competition is always healthy so long as nobody's playing dirty pool.
You are on crack. Seriously cheap crack. There are so many problems with your idea I can't even figure out where to start.
One badass file that everything clamours to get access to, even through a well defined API is going to be hell. Contention abounds. A bug can kill the entire system configuration. And then you throw the saviour of the digital world into it: XML. Why don't you just make it a binary file and have every single problem the Win32 registry has.
A unified configuration scheme is a great idea. XML really isn't IMO the solution, and one big badass file is certainly the wrong way to go about it.
I didn't say I couldn't create my own schema. I said that something as obvious as what I described should already exist somewhere, and I was hoping someone reading knew where it was. I didn't want to reinvent the wheel with something inferior. I mean I can't be the only person on the planet who is looking for even 70% of what I listed. I'm sure it's been done before and likely includes things I didn't even think of.
That's the bigger problem with LDAP I think -- once your schema's designed it's a bear to change to something else. Possible yes, but pretty... certainly not.:-)
My biggest question is in schema design. I mean there are hundreds of schemas out there yet it seems next to impossible to put them together intelligently to get an object that would allow something along the lines of
name, nickname,position, etc.
contact info (address,phone,email)[home/biz/alternate]
samba/u nix/whatever identities
GPG public/private keys, x.509 certificate information, etc.
connection information (secretaries, wife, children, pets)
anniversaries (birthday,marriage,etc.)
photo
biometric ID
some access to a free-form series of notes on the person
The last one could be done with some kind of GUID you could link outside of the LDAP server but the rest seem to be impossible to bring together in an organized fashion. I mean this kind of thing must have been done ages ago but the schemas seem to be lost. I'd love to get an LDAP server set up to keep this information in one place but I always get lost in the schema design.
Ext3 (or resiser, or xfs, or any other jfs for that matter) just helps to ensure that your metadata's consistent. They do this by making metadata or filesystem state writes atomic (I think that's the term) -- basically when the kernel updates a file, the metadata is written to disk right away, but the actual file data is still in the write cache.
Now there are all kinds of tiny variants to this, and ext3 *can* write your file data to the journal too, but it makes the system quite a bit slower.
Supercaps are totally useless for anything but low current discharge -- sure you can pack 5F into a tiny space but the energy practically dribbles out of it. They're *great* for backup batteries for things like your Palm, though.
WTF are you talking about? A "fuse [that] doesn't work" is one that's already burned out. There is no such thing as a "defective" fuse that lets way too much current through. Fuses are safety devices. Fuse manufacturers are very careful for reasons of liability. Your "what if" is as irrelevant as "what if tires were TRIANGULAR?"
Nothing is infallible, and I'm willing to put money on there being at least a half dozen lawsuits over fuses which failed to protect their load, and not through misapplication. Fusing is big business (I work in the power electronics field) and while they're a hell of a lot more reliable than breakers, they're not perfect.
Remember that fuses are like MOVs in that respect -- you can't test them without damaging them.
Moore's Law has absolutely nothing to do with the speed of integrated circuits, it is talking about the complexity of the designs doubling roughly every 18 months. Complexity doesn't necessarily mean speed.
tried almost all other distros but in the end its always slack that stays
Amen. Amen, amen, amen!
I too have tried practically every other distro out there and even a few BSDs but I always come back to Slack. It's strange, but all the other's package managers ever seem to do is piss me off by thinking they know best, or by installing a bazillion things that I don't need because the dumbass maintainer doesn't have a -full and -lite version of a package. Same with the "helpful" configuration interfaces... gimme Slack with a SysV init (yeah I am balking the trend there a little but it's *so* much nicer than editing a script whenever I want to alter the order or install new software) and one of the new package updaters like Swaret or slapt-get and I'm quite happy.
Speaking of swaret -- everyone in freenode's #slackware says it'll eat my children eventually -- odd, but it seems to just work for me... is this just old prejudices or what?
And I don't think Microsoft is really afraid of Open Office... I tried using Open Office, and after a day of trying to figure out how to do what I did in MSOffice, I just went back to using MSOffice. It does what I need it to do and how I need it done.
That's funny, because I prefer OO to MSO now for many things... the only thing I can't stand about OO Calc is how if I select an area and hit ^X for cut, the cursor stays at the bottom right of the selected block instead of the top left... I mean if I must want to move a block over a few places it's a royal pain in the ass with OO. But the other things -- ESPECIALLY database access -- are so much nicer with OO.
Poor standby battery life (less than 8h), poor call quality, poor buttons, craptastic configuration web interface, Zero support from Pulver, weak range, weak processor (no WEP for you)... crappy stand for charging (no positive "mate"), crappy display... what else?
I've been using Slackware since it diverged from SLS so many years ago and I'd have to say that you are 100% correct about Debian... nice idea but if I wanted politics, I'd tune in to CSPAN. Their rabid use of GNU/everything has utterly and totally turned me off of the distribution. Gentoo? I don't think so. I think that I'd move to something like Vector Linux which is Slackware based and has a more sensible set of installer defaults (I use my own tagsets though already). The only real reason I'm not using Vector currently is because the ASCII art penguin has to go for starters but more importantly, Slackware is still the 'root' of it and why go with a little offshoot that doesn't change much at the current time?
We're using The Vexi Project to design our customer service, sales and call tracking system (essentially the core of our office systems). We are doing this instead of relying on technologies centered around existing systems because, frankly, all you get with COTS is a shortcut to dependence on a disinterested third party.
We're not Ford. We haven't got the pull of someone like Microsoft. We have been bitten in the past by COTS software (oh, AutoCAD, AccPAC, Microsoft for starters) -- support contracts went up in price every single year and the problems/bugs resolved were trivial and inconsequential while the larger problems were glossed over or promised "in the next version." And then of course you get to the beloved statement of "We're end-of-lifeing the product. Sorry about your luck."
With OSS this just does not happen. In fact, I would go so far as to say it's impossible. With Vexi, the engine is open source and runs on multiple platforms (Java and then native Win32, Linux and OSX). We didn't go the DHTML route because now you've tied yourself to a browser. We've hired a contractor (in fact, one of the driving forces behind Vexi) to continue developing the widget set and work away at the engine while writing exactly what we want and in the end, it is ours. We can try to make it a COTS system and sell it (not likely), we can expand it either by writing the extensions ourselves or by hiring him again (or anyone else) and -- here's the best part -- nobody can take it from us. We don't have to worry about retooling when the company gets bought out (SSH Sentinel) and, as I've said before, as our business grows and changes so can our software.
You simply cannot get that with COTS. We've been there and done that and in my opinion it's a very bad thing to do -- to put your core business flow into someone else's hands. No two businesses run quite the same way and this is the exact reason why there are a hundred different COTS systems trying to be "the one."
Well I already said that it's almost every time that I have weird-ass dreams that there's solar activity going on, not every time. Once in a while I'll check the sites and yes... no memorable dreams generally means little/low solar activity.
As they say, correlation does not mean causation, I'm just giving a datapoint. No need to act all superior.
I don't know about you but I have been tracking the weird-ass dreams that I remember for the last year or so and whenever I have a "rash" of memorable dreams I check the various sun tracking sites and almost without fail I will find that there was some kind of burst of solar activity going on around that time. It could just be coincidence but it's kind of spooky that I can track these two events more or less in lock-step.
I'm looking all over that redhat.com/beta link you posted but cannot see a single thing relating to a new beta for bugzilla with postgres. Are you just pulling my leg or what's going on?:-)
That google image link has Google textads for ebay... "Airbus A380 for sale. aff Check out the deals now! www.eBay.com" and "Low Priced Airbus A380 Huge Selection! (aff) ebay.ca"
DCOM is literally a reverse engineered DCE-RCP, to the point where it is wire compatible with it.
...
but once compiled, a DCE rcp client/server can talk to a DCOM client/server, assuming you are not trying to use any of the security built into the DCE-RPC
Do you have more information? I am very interested in learning how to interop these two technologies! (my email addy is a valid one)
Slashdot Mirror
You gotta be fucking kidding me.
MySQL's the fast but clumsy child, whereas Postgres is the slow but methodical child. MySQL tacks on features and calls them stable yet you can't trust your data to it. Postgres is ver conservative out of the box but each release includes significant speed increases.
MySQL is very fast for super-simple selects and inserts. Throw anything more at it and it crawls. Postgres, on the other hand, is pretty conservative in terms of memory usage but once you increase the shared memory allocation and turn off the forced-fsync it handles much faster than MySQL for even slightly complex queries and will outstroke MySQL handily as you pile on connections and complex queries.
MySQL's not had the features because a) they slow the DB down and b) their entire philosophy is "we just store data. You make sure it's good before you give it to us or we'll make it good in our own way and not tell you."
MySQL 5 is a HUGE leap forward for MySQL and most of the points will probably become moot. Let's just hope they fix the default-value fiasco also...
Are they planning on fixing the artistic license that MySQL routinely takes with the data thrown at it? I'm talking about autotruncation, auto "converting" strings to numbers, NULL and 0 being the same thing... Until MySQL takes data consistency or at least validation seriously it will never make it in this shop.
When MySQL 5 is released PostgreSQL will get some more Open Source competition and that is a good thing.
Totally agreed. I'm a fan of Postgres, this is true, but competition is always healthy so long as nobody's playing dirty pool.
Story
You are on crack. Seriously cheap crack. There are so many problems with your idea I can't even figure out where to start.
One badass file that everything clamours to get access to, even through a well defined API is going to be hell. Contention abounds. A bug can kill the entire system configuration. And then you throw the saviour of the digital world into it: XML. Why don't you just make it a binary file and have every single problem the Win32 registry has.
A unified configuration scheme is a great idea. XML really isn't IMO the solution, and one big badass file is certainly the wrong way to go about it.
You missed my point, almost entirely.
I didn't say I couldn't create my own schema. I said that something as obvious as what I described should already exist somewhere, and I was hoping someone reading knew where it was. I didn't want to reinvent the wheel with something inferior. I mean I can't be the only person on the planet who is looking for even 70% of what I listed. I'm sure it's been done before and likely includes things I didn't even think of.
That's the bigger problem with LDAP I think -- once your schema's designed it's a bear to change to something else. Possible yes, but pretty... certainly not. :-)
My biggest question is in schema design. I mean there are hundreds of schemas out there yet it seems next to impossible to put them together intelligently to get an object that would allow something along the lines of
The last one could be done with some kind of GUID you could link outside of the LDAP server but the rest seem to be impossible to bring together in an organized fashion. I mean this kind of thing must have been done ages ago but the schemas seem to be lost. I'd love to get an LDAP server set up to keep this information in one place but I always get lost in the schema design.
There aren't many... as in there aren't any!
You're most likely mistaken.
Ext3 (or resiser, or xfs, or any other jfs for that matter) just helps to ensure that your metadata's consistent. They do this by making metadata or filesystem state writes atomic (I think that's the term) -- basically when the kernel updates a file, the metadata is written to disk right away, but the actual file data is still in the write cache.
Now there are all kinds of tiny variants to this, and ext3 *can* write your file data to the journal too, but it makes the system quite a bit slower.
you're actually working 168 hours a week.
Huh? Four 12 hour days a week is 48 hours. Even if you add in the 3h roundtrip commute that's 60 hours a week.
I thought it was common knowlege that ants communicated through scent trails?
Supercaps are totally useless for anything but low current discharge -- sure you can pack 5F into a tiny space but the energy practically dribbles out of it. They're *great* for backup batteries for things like your Palm, though.
WTF are you talking about? A "fuse [that] doesn't work" is one that's already burned out. There is no such thing as a "defective" fuse that lets way too much current through. Fuses are safety devices. Fuse manufacturers are very careful for reasons of liability. Your "what if" is as irrelevant as "what if tires were TRIANGULAR?"
Nothing is infallible, and I'm willing to put money on there being at least a half dozen lawsuits over fuses which failed to protect their load, and not through misapplication. Fusing is big business (I work in the power electronics field) and while they're a hell of a lot more reliable than breakers, they're not perfect.
Remember that fuses are like MOVs in that respect -- you can't test them without damaging them.
How can a single 7805 rated for a maximum Icc of 1A provide the couple dozen amperes to provide even the output equivalent to an Easy-bake oven?
It's called a pass transistor. 2n3055 or equivalent and, as a bonus, you can use the 2n3055 to provide extra heat!
Moore's Law has absolutely nothing to do with the speed of integrated circuits, it is talking about the complexity of the designs doubling roughly every 18 months. Complexity doesn't necessarily mean speed.
tried almost all other distros but in the end its always slack that stays
Amen. Amen, amen, amen!
I too have tried practically every other distro out there and even a few BSDs but I always come back to Slack. It's strange, but all the other's package managers ever seem to do is piss me off by thinking they know best, or by installing a bazillion things that I don't need because the dumbass maintainer doesn't have a -full and -lite version of a package. Same with the "helpful" configuration interfaces... gimme Slack with a SysV init (yeah I am balking the trend there a little but it's *so* much nicer than editing a script whenever I want to alter the order or install new software) and one of the new package updaters like Swaret or slapt-get and I'm quite happy.
Speaking of swaret -- everyone in freenode's #slackware says it'll eat my children eventually -- odd, but it seems to just work for me... is this just old prejudices or what?
No .iso's to download? Another strike. Ya, I'm a free beer kinda guy.
Acutally I downloaded the SuSE 9.2 pro .iso the other day and installed it. I'm a slackware user myself, but SuSE is *nice*.
And I don't think Microsoft is really afraid of Open Office... I tried using Open Office, and after a day of trying to figure out how to do what I did in MSOffice, I just went back to using MSOffice. It does what I need it to do and how I need it done.
That's funny, because I prefer OO to MSO now for many things... the only thing I can't stand about OO Calc is how if I select an area and hit ^X for cut, the cursor stays at the bottom right of the selected block instead of the top left... I mean if I must want to move a block over a few places it's a royal pain in the ass with OO. But the other things -- ESPECIALLY database access -- are so much nicer with OO.
The WiSIP sucks rocks, dude.
Poor standby battery life (less than 8h), poor call quality, poor buttons, craptastic configuration web interface, Zero support from Pulver, weak range, weak processor (no WEP for you)... crappy stand for charging (no positive "mate"), crappy display... what else?
I've been using Slackware since it diverged from SLS so many years ago and I'd have to say that you are 100% correct about Debian... nice idea but if I wanted politics, I'd tune in to CSPAN. Their rabid use of GNU/everything has utterly and totally turned me off of the distribution. Gentoo? I don't think so. I think that I'd move to something like Vector Linux which is Slackware based and has a more sensible set of installer defaults (I use my own tagsets though already). The only real reason I'm not using Vector currently is because the ASCII art penguin has to go for starters but more importantly, Slackware is still the 'root' of it and why go with a little offshoot that doesn't change much at the current time?
We're using The Vexi Project to design our customer service, sales and call tracking system (essentially the core of our office systems). We are doing this instead of relying on technologies centered around existing systems because, frankly, all you get with COTS is a shortcut to dependence on a disinterested third party.
We're not Ford. We haven't got the pull of someone like Microsoft. We have been bitten in the past by COTS software (oh, AutoCAD, AccPAC, Microsoft for starters) -- support contracts went up in price every single year and the problems/bugs resolved were trivial and inconsequential while the larger problems were glossed over or promised "in the next version." And then of course you get to the beloved statement of "We're end-of-lifeing the product. Sorry about your luck."
With OSS this just does not happen. In fact, I would go so far as to say it's impossible. With Vexi, the engine is open source and runs on multiple platforms (Java and then native Win32, Linux and OSX). We didn't go the DHTML route because now you've tied yourself to a browser. We've hired a contractor (in fact, one of the driving forces behind Vexi) to continue developing the widget set and work away at the engine while writing exactly what we want and in the end, it is ours. We can try to make it a COTS system and sell it (not likely), we can expand it either by writing the extensions ourselves or by hiring him again (or anyone else) and -- here's the best part -- nobody can take it from us. We don't have to worry about retooling when the company gets bought out (SSH Sentinel) and, as I've said before, as our business grows and changes so can our software.
You simply cannot get that with COTS. We've been there and done that and in my opinion it's a very bad thing to do -- to put your core business flow into someone else's hands. No two businesses run quite the same way and this is the exact reason why there are a hundred different COTS systems trying to be "the one."
Well I already said that it's almost every time that I have weird-ass dreams that there's solar activity going on, not every time. Once in a while I'll check the sites and yes... no memorable dreams generally means little/low solar activity.
As they say, correlation does not mean causation, I'm just giving a datapoint. No need to act all superior.
I don't know about you but I have been tracking the weird-ass dreams that I remember for the last year or so and whenever I have a "rash" of memorable dreams I check the various sun tracking sites and almost without fail I will find that there was some kind of burst of solar activity going on around that time. It could just be coincidence but it's kind of spooky that I can track these two events more or less in lock-step.
I'm looking all over that redhat.com/beta link you posted but cannot see a single thing relating to a new beta for bugzilla with postgres. Are you just pulling my leg or what's going on? :-)
That google image link has Google textads for ebay... "Airbus A380 for sale. aff Check out the deals now! www.eBay.com" and "Low Priced Airbus A380 Huge Selection! (aff) ebay.ca"
Personally I find that frickin' hilarious.
DCOM is literally a reverse engineered DCE-RCP, to the point where it is wire compatible with it.
...
but once compiled, a DCE rcp client/server can talk to a DCOM client/server, assuming you are not trying to use any of the security built into the DCE-RPC
Do you have more information? I am very interested in learning how to interop these two technologies! (my email addy is a valid one)