FF3 is almost infinitely better than 1.5 and 2 in terms of performance, stability, and memory usage. However, there are still some niggling performance issues that make me tear my hair out. Still, from someone who is most definitely NOT a FF fanboi, it's actually their best release by far and worth checking out.
Indeed it's not acceptable to crash on bad input; my comment about W3C standards was in reference to his whinging about having to write special code around IE, when actually he should be writing that code anyway to be standards compliant (and for usability). Also, if indeed it was fixed in IE 7 (that is to say it was broken in IE 6) then I guess that answers his question of "Do they plan to fix the select bug?".
I'm quite aware of the...quirky nature of the select widget in IE. Back in the day I had to work around the z-index problems by drawing an iframe "undercoat" underneath any element that might get drawn over a select box in a DHTML heavy app. Frankly ridiculous.
WTF? It certainly doesn't do that on my IE 7 installation. Also, your "custom code" complaint doesn't hold much water as the W3C specs state that a select element must contain at least one option element.
No. In order to rename the file remotely you already need root. And even ignoring that, you would still need physical access to use the newly exploited shell.
Your comment is akin to saying "Ah, but what if someone finds a way to remotely append init=/bin/bash to Grub?" There's no weakness in Linux there, as you'd need to have root on the box in order to do such a thing, and then after the shutdown -r you'd be fucked anyway as it sat at a shell 1000 miles away waiting for someone to type into the console.
PC sales growth has slowed in the U.S. as consumers and businesses curb purchases to cope with economic struggles. That has left Microsoft with a greater proportion of sales in developing countries, where the programs sell at lower prices.
``Not only are they charging less, but the highest growth rates in terms of new PC sales are in emerging markets,'' said Gartner Inc. analyst Neil MacDonald. ``They have to do this. The risk is that Linux gets a foothold and then affects their long- term revenue more significantly.''
So the huge decrease in sales revenue is equated with the decrease in sale price, not sales numbers.
Just one nitpick - stored procedures are not always faster, and for much the same reason: as they're pre-parsed and planned (which is the time-saving part) the query planner can make no use of the actual parameters when deciding the optimal plan. I have found (admittedly rarely) that turning a literal query into a stored procedure can increase the run time of the query several times over.
Honestly, no - I've never used Hibernate in particular. However, my job involves not only the usual simple persistance/retrieval stuff (which ORM makes easy) but also writing complex reports (or even not-so-complex reports that just really don't fit the OO view of the data).
As you point out, a good ORM framework will allow you to query this anyway, but the problem there is you're not reading my posts in the context of the OP. I'm not denying ORM has its uses (that would be retarded; by definition it's needed in some form if you want to use an OO program with a relational backend) but the OP was asking why anyone would access a DB with anything other than an ORM framework - and the answer is that sometimes the query you run just doesn't cleanly fit the OO model you've constructed.
Your pragmatic comment that "you can use native SQL" rather negates the OP's point. If you're using native SQL then the fact you're using Hibernate is irrelevant; you could be using any DB interface that supports prepared statements (or otherwise parametised queries) and there I would have to agree that in virtually all cases you really should.
Actually, the more I think about it, the more I suspect we're not even at odds here. I've nothing against using ORM frameworks to interface between OO code and an RDBMS, and your comment about HQL/SQL suggests you've nothing against leaving the application-specific object model in order to retrieve datasets that don't fit it. My argument is with this strange dogmatism that seems too common at the moment, that OO is always good and everything else is always bad. Once again, the concept of "right tools for the job" is seen as heresey.
No, you're right - it combines the worst aspects of both, making something that looks superficially like an ODBMS to the programmer, whilst having none of the features of an ODBMS to make it perform well.
All I care about is the quality of the songs - ironically, this is exactly what "it's all about the music" types don't do when they're complaining about who wrote songs vs who sang them.
I am well aware that Will didn't write "Leave right now", and that Girls Aloud didn't write "The Show". However, I'm also aware that Eg White and Brian Higgins are great songwriters, while the aforementioned singers are great performers/vocalists. Higgins even pointed out that as he is a terrible singer, what exactly should he be doing other than writing songs for other people? If you've got a great songwriter and a great singer, what's wrong with pairing them up?
Don't give me the whole pitch correction spiel either - that's an issue with modern production pretty much everywhere, just like the loudness wars. Also, the genesis of these acts means that we have a rather unique record of exactly how they sound uncorrected and they are good singers. No, they're not as technically good as many classical singers, but they are good pop singers which is a quite different beast (side note: Charlotte Church was at best an OK pop singer).
I'm sure Michelle is just lovely, and she has a reasonable voice, but can you honestly say she had any star potential? Obviously her debut track was just dross, but I really can't see her having fared any better if she'd been given Bleeding Love.
Interesting you bring up Pop Idol, because that sort of reality show has brought some very good acts/songs. Will Young has released great tracks, both the UK and USA seem to love Leona Lewis (I don't know many of her songs, but Bleeding love is ace) and Girls Aloud are just fucking awesome through and through. Some of the output has been pure shite, of course - Michelle "sympathy vote" McManus springs to mind.
Yes, I'm a big pop music fan. No, that doesn't make my opinion less valid than yours. Be careful not to conflate what you subjectively dislike with what is objectively bad. There is good and bad home-recorded music, good and bad indie-label music, and good and bad big-label manufactured music. This Slashdot-favourite idea that any music other than The Who or a singer/songwriter fingerpicking an acoustic into his portastudio must be shit is both arrogant and narrow-minded.
Does that $800 studio in a box include the U87, good preamps and eqs, skilled recording, mixing and mastering engineers?
Yes, making music yourself is easier than ever, and the results better quality than ever. But claiming your cheap digital multitrack produces better results than studio productions of a decade ago is frankly foolish.
What? The Word thing was kind of a joke. I have no interest in Microsoft, and am in fact perfectly happy with their software. However:
The only relevence[sic] these documents have now is that they expose the bank's ultra-rich clients suspiciously funneling money through Cayman Islands trusts nearly a decade ago.
What part of that makes publishing them OK? The fact that it happened "nearly a decade ago"?
Meanwhile, WikiLeaks has responded harshly to the recent statement issued by the bank Julius Baer.
Am I the only one who read that "harsh response" and heard little more than "we were treated unfairly in court!"? Especially poor was the section that could be paraphrased as "They said we posted confidential bank records, but they weren't - they were Word files! And quite old! And maybe just a bit confidential! Did we mention Word?"
But not all windows can communicate with each-other - indeed I'd say the vast majority can't. If window 1 doesn't have a handle to window 2 and vice versa - either directly or through intermediate DOM elements - then I think I'm right in saying that the DOMs will never be able to communicate with each-other. There's no way to dynamically obtain such a handle after the window has been created (correct me if I'm missing something).
As such, sets of unrelated windows (which will usually consist of a single window) can run in their execution thread. Furthermore, the code to do this graph traversal is pretty much already implemented in the mark and sweep garbage collector!
While I mostly agree with you about MD5 being a poor choice, it is not completely useless if implemented carefully using a nonce.
Mostly I posted this because I like using the word 'nonce'. The correct answer, of course, is to use SSL. If you don't want to pay for a signed certificate then use a self-signed cert. If you think the security warnings will put people off then default them to an HTTP login, with a link to HTTPS for the paranoid.
But if you actually read the full thread, you'll find that the various solutions are to boot into safe mode, to use the recovery console, and to revert to a previous configuration. The redundancy is there, whether you know about it or not.
Thankfully now in Cambridge, and the youths are a lot more sensible and have several parks in which to hang out away from people.
I think you've inadvertently managed to say something quite insightful there, despite your previous comment about wanting these devices. The fact is, apart from the genuine sociopaths, nobody says "Hey, tonight let's go hang around a bus stop and terrorise people while acting like thugs". It's more a case of "Where can we go hang out other than one of our parents' houses?"
Nowadays, we've got our own houses to socialise in. In our late teens, we had pubs to socialise in. In early-mid teens...not so much choice. It's either a park, or outside the shops.
Slashdot Mirror
Which is why you don't view your encrypted child porn on your work computer. But you may want to view your encrypted work data on that same machine.
Yes, but be sure to also check out the bits where he says "but don't do it like this - if you don't try and emulate classical OO it's a lot easier".
FF3 is almost infinitely better than 1.5 and 2 in terms of performance, stability, and memory usage. However, there are still some niggling performance issues that make me tear my hair out. Still, from someone who is most definitely NOT a FF fanboi, it's actually their best release by far and worth checking out.
I'm quite aware of the...quirky nature of the select widget in IE. Back in the day I had to work around the z-index problems by drawing an iframe "undercoat" underneath any element that might get drawn over a select box in a DHTML heavy app. Frankly ridiculous.
WTF? It certainly doesn't do that on my IE 7 installation. Also, your "custom code" complaint doesn't hold much water as the W3C specs state that a select element must contain at least one option element.
Your comment is akin to saying "Ah, but what if someone finds a way to remotely append init=/bin/bash to Grub?" There's no weakness in Linux there, as you'd need to have root on the box in order to do such a thing, and then after the shutdown -r you'd be fucked anyway as it sat at a shell 1000 miles away waiting for someone to type into the console.
Just one nitpick - stored procedures are not always faster, and for much the same reason: as they're pre-parsed and planned (which is the time-saving part) the query planner can make no use of the actual parameters when deciding the optimal plan. I have found (admittedly rarely) that turning a literal query into a stored procedure can increase the run time of the query several times over.
As you point out, a good ORM framework will allow you to query this anyway, but the problem there is you're not reading my posts in the context of the OP. I'm not denying ORM has its uses (that would be retarded; by definition it's needed in some form if you want to use an OO program with a relational backend) but the OP was asking why anyone would access a DB with anything other than an ORM framework - and the answer is that sometimes the query you run just doesn't cleanly fit the OO model you've constructed.
Your pragmatic comment that "you can use native SQL" rather negates the OP's point. If you're using native SQL then the fact you're using Hibernate is irrelevant; you could be using any DB interface that supports prepared statements (or otherwise parametised queries) and there I would have to agree that in virtually all cases you really should.
Actually, the more I think about it, the more I suspect we're not even at odds here. I've nothing against using ORM frameworks to interface between OO code and an RDBMS, and your comment about HQL/SQL suggests you've nothing against leaving the application-specific object model in order to retrieve datasets that don't fit it. My argument is with this strange dogmatism that seems too common at the moment, that OO is always good and everything else is always bad. Once again, the concept of "right tools for the job" is seen as heresey.
No, you're right - it combines the worst aspects of both, making something that looks superficially like an ODBMS to the programmer, whilst having none of the features of an ODBMS to make it perform well.
Because not everyone has swallowed the fashionable BS about ORM being the way ahead, and instead are happy with a pure relational model?
Batten down the handles - this teacup's in for a stormy night!
I am well aware that Will didn't write "Leave right now", and that Girls Aloud didn't write "The Show". However, I'm also aware that Eg White and Brian Higgins are great songwriters, while the aforementioned singers are great performers/vocalists. Higgins even pointed out that as he is a terrible singer, what exactly should he be doing other than writing songs for other people? If you've got a great songwriter and a great singer, what's wrong with pairing them up?
Don't give me the whole pitch correction spiel either - that's an issue with modern production pretty much everywhere, just like the loudness wars. Also, the genesis of these acts means that we have a rather unique record of exactly how they sound uncorrected and they are good singers. No, they're not as technically good as many classical singers, but they are good pop singers which is a quite different beast (side note: Charlotte Church was at best an OK pop singer).
I'm sure Michelle is just lovely, and she has a reasonable voice, but can you honestly say she had any star potential? Obviously her debut track was just dross, but I really can't see her having fared any better if she'd been given Bleeding Love.
Yes, I'm a big pop music fan. No, that doesn't make my opinion less valid than yours. Be careful not to conflate what you subjectively dislike with what is objectively bad. There is good and bad home-recorded music, good and bad indie-label music, and good and bad big-label manufactured music. This Slashdot-favourite idea that any music other than The Who or a singer/songwriter fingerpicking an acoustic into his portastudio must be shit is both arrogant and narrow-minded.
Yes, making music yourself is easier than ever, and the results better quality than ever. But claiming your cheap digital multitrack produces better results than studio productions of a decade ago is frankly foolish.
You Americans crack me up.
In order to make huge amounts of money, artists should charge huge amounts for their music.
As such, sets of unrelated windows (which will usually consist of a single window) can run in their execution thread. Furthermore, the code to do this graph traversal is pretty much already implemented in the mark and sweep garbage collector!
Mostly I posted this because I like using the word 'nonce'. The correct answer, of course, is to use SSL. If you don't want to pay for a signed certificate then use a self-signed cert. If you think the security warnings will put people off then default them to an HTTP login, with a link to HTTPS for the paranoid.
I agree. He is roughly as historically important as the entire Police Academy series.
But if you actually read the full thread, you'll find that the various solutions are to boot into safe mode, to use the recovery console, and to revert to a previous configuration. The redundancy is there, whether you know about it or not.
I think you've inadvertently managed to say something quite insightful there, despite your previous comment about wanting these devices. The fact is, apart from the genuine sociopaths, nobody says "Hey, tonight let's go hang around a bus stop and terrorise people while acting like thugs". It's more a case of "Where can we go hang out other than one of our parents' houses?"
Nowadays, we've got our own houses to socialise in. In our late teens, we had pubs to socialise in. In early-mid teens...not so much choice. It's either a park, or outside the shops.