While it is no particular advantage to the open source movement to have open source tools on windows, it is of extreme benefit to have platform independent tools in use on Windows. Open source benefits substantially from the availability of non-proprietary software on the (proprietary) OS with the largest market share. The idea that Open Source advocates should be against open source programs on Windows is asinine. Is anyone really worried that Windows will transpire to be a superior OS to run platform independent software?
I'm not particularly enamoured with PDAs - particularly the ones which also try to be a phone. I have, however been trying to research my options for a device more portable than my laptop on which I can easily read and reply to email; use the web etc. It would need to have the handwriting recognition facilities of the best modern PDAs; have a substantial battery life - at least 8 hours; and finally, a c. 7" screen with at least SVGA resolution. I wouldn't mind if it was fairly heavy but I would find anything larger than c. A5 as cumbersome as a laptop when used on my lap in an sumptuous arm chair. I'm sure there's a market for such a device but I can't find one... I did find these - but while this is the format I want, this product is only a DVD player.
I wonder - am I alone in seeing the "Google Groups 2" as a significant improvement on the original?
I like the improved 'posting' speed; I love the 'starred topics' (Though I remain sceptical that the 'new posts' feature works properly - I keep thinking "new since when?"). I like the idea that a thread has become the notional unit searched in the new UI - Google Groups 2 far better suits my needs.
... to provide evidence for low energy nuclear reactions. These experiments involved low energy deuterium beams impinging on deuterium loaded metal foils such as titanium.
In moments like these I'm glad I bought the tin foil hat and not the more luxurious titanium one.
Real cold fusion nut jobs would choose a palladium hat.
My own camera phone is on-order... so I can't say for sure... but I strongly suspect that the screen will prove inadequate for any complex text - I expect a phone number, street name or some such would be legible (from what I've seen on others' phones) - though for my own use I'm not so bothered - I can easily display the images on my desktop/laptop when I'm at home or in the office... which (luckily) happens to be when I find I wish I remembered such details.
I used to think exactly the same. I was given an SLR camera 20 years ago and I probably took a hundred or so photos with it as a teenager. 3 years ago I bought a 2MP point and shoot digital - and I've taken around 30 to 40 photos so far - and kept 2 or 3 (as digital images.) I'm no photographer - most of my digital photos were to document things - I bought the digital to document where someone had driven into my car then driven off. I thought "another camera - how could that possibly be valuable to me?"
Keeping that thought in mind - I've found dozens of occasions I'd have made use of one... just not for taking snaps of people, or places. I frequently find myself noticing a sign with a phone number or a closed shop or restaurant with opening hours posted in the window. I never seem to have a pen or paper (and copying down a list of opening hours would be very tedious anyway) - but a photo of the sign taken using my mobile would be perfect... it would take only a couple of seconds, and I wouldn't loose the information (unless I loose my phone.)
Plausible... but that doesn't explain why it is better that I have to blindly fumble for a working firmware where an OSS project could simply give me an MD5 to verify.
I can't argue with your pragmatism - and I guess I'd be far more inclined to take the same route if I started again now... like yourself I followed the handbook, then after that failed I tried other how-to documents... Ideally I'd like to see the handbook corrected so as not to misguide anyone else. I understand the concerns about performance - but to be honest performance isn't an issue for me... A reliable always-on cheap 28800 baud link would suit my needs...
I might have asked exactly the same question on Sunday when, after much effort I got as far as seeing/var/log/messages reporting "Where is the crappy modem?" (I'm not embellishing the colourful language!)
In an attempt to explain my reasoning (which one might reasonably consider flawed) I decided to by a Speedtouch USB ADSL modem rather than an Ethernet one for several reasons:
I want to keep my LAN physically separate from internet traffic - I wanted to use BSD as a secure gateway to the net for my insecure LAN - I didn't particularly want to install 2 Ethernet cards for the server PC either - and an entire router connected to only one host seems pointless.
According to the BSD handbook the Alcatel (AKA Thompson) "Speedtouch USB" ADSL modem was supported. Only later did I realise they were talking about a specific Speedtouch model (and not any Speedtouch with a USB interface.)
The modem was used initially from a Windows box - and the USB modem performed wonderfully in that configuration.
So far I've discovered that the "Alcatel Speedtouch USB" modem isn't the same as a "Speedtouch 330 USB" modem - and, of the latter, the silver ones are "Rev 4" and need different firmware to that documented in most howto documents. I've not established why I can't establish a PPPoA connection using it.... Nor why the FreeBSD handbook seems to assume I'm using an Ethernet device (judging from the sample configuration) right under a heading specifying USB. I'd love to hear from someone who has a Speedtouch 330 Rev4 modem working using PPPoA either on Free/Open BSD... especially if they will share the specific details of their working configuration... all the documents I can find seem to be contradictory or assume that my USB modem transmogrifies into an IP addressable device.
It is only a for-home system... the Speedtouch modem works fantastically under windows - and hell - all I want it to do is chuck packets at my ISP from my BSD box. As the handbook suggests just such a configuration - I'm sure you'll agree it shouldn't be a big problem to set up. I'd prefer not to use a "wireless router" as I want to use more advanced packet filtering and IPSEC encryptions which are easy with BSD. I like the conceptual simplicty of 1 ADSL adaptor; 1 Ethernet card; 1 wireless card. In order to achieve a comparable level of security but avoiding dedicated ADSL modems - I'd need a second ethernet card and an ADSL router (from which I'd use precious little functionality) as I really want my LAN insulated from the ADSL link by a gateway machine. My budget isn't all that shoestring - but I'm loathed to spash out on new kit when I already own something that works perfectly well for my purposes. Obviously a commercial solution would be to buy in a working solution... but this isn't a commercial project.
I can't fault the preference for redistributable firmware - that's obvious. However, we don't need any third party to change its behaviour to overcome my hassles... just clear, accurate descriptions (with MD5 sums where using 3rd party components) of the components themselves and their configuration. Downloading drivers is no hassle for me - and, I suspect, if the drivers worked painlessly and proved easily configured then the manufacturer would be happy to put the firmware on the CD distributed with the hardware. Redistributable would be best but I'd be happiest if we addressed the OSS system first in order that the only problem is redistribution of the firmware (which seems a relatively minor inconvenience right now.)
An amusing chicken and egg quandary for sure - but not a serious concern for most BSD/Linux users who usually have at least occasional access to another internet connected PC. I'd say misleading or missing documentation is a bigger issue - I certainly can't download firmware if I don't know which firmware I should be downloading!
Hmmm - that makes Atheros looks just as supported as in FreeBSD... My concerns (about a year ago) were that the only references to ATH in OpenBSD were messages suggesting that it wasn't supported - I didn't feel inclined to install it to check the manual pages. I don't suppose you can also tell me that OpenBSD can also solve my problems with my "Speedtouch 330 (Rev 4)" USB ADSL modem assuming an ISP demanding PPPoA? If I can solve my headaches with a change from FreeBSD to OpenBSD, I'd do that in an instant.
Part of me wants to back Theo arguing for distributable firmware - but another part of me feels that there is still a lot that can be achieved without requiring any re-licensing.
I'm currently stuck trying to get my Alcatel/Thompson "Speedtouch 330" (Revision 4) ADSL modem to work under FreeBSD 5.3. Downloading the 'firmware' was a pain but much of that could have been resolved with some good documentation and an MD5 to verify the correct version. Even now I have the device recognised following the handbook doesn't get me connected... and offers precious little information about how to make appropriate configuration.
I suppose the response might be that that OpenBSD would do this fine - though I chose FreeBSD as a result about concerns about OpenBSD support for the Atheros chipset in my Dlink DWL G520 PCI wireless net card (which is straightforward to configure in FreeBSD.) Aaaagh!
I'm compelled to reply - your rant is off-base on at least a couple of points - though I concede that you've identified an interesting mentality. Here's my own rant!
Let me start by explaining how I value goods: I start by making sure I compare like-with-like, then arrive at a value of the item itself - this is the cheapest price at which I can buy the item. Next I factor in convenience (which, to me at least) is extremely important - and finally assess any risks (such as a risk of receiving faulty goods and any cost or effort involved in resolving the issue.) I compare my weighted prices and choose the cheapest (weighted) price - it is the "sane" way.
Things are more complicated when considering dissimilar products or (frequently) re-branded products at different prices. I find re-branding extremely counter productive... such a product may introduce compatibility concerns or be more difficult to repair or service - hence I assess risk adversely. Expanding the namespace is hardly convenient either. For example, I live in the UK, and 2 of the main electrical retailers are Currys and Comet, and it is normal to find both stores within easy walking distance of each other - they usually appear to be locked in a price war. Interestingly both stores are owned by the same group that owns Dixons (another large UK electrical retailer.) The two store names are, in my opinion, a scam to present a false impression of competition. I still buy from these stores but I have become extremely price and brand sensitive about their offers. When their sales-operatives enter their sales spiel for extended-warranty I reply with my own - asking if they are knowingly supplying defective goods; and ask if they intend to comply with UK consumer protection laws. The conversation usually ends with - if the goods turn out to be faulty after a year I will simply remember to replace with a product by a different manufacturer from a different retailer.
I _DESPISE_ money back offers - to me they scream "FRAUDULET VENDOR ALERT!" - so... I guess they are useful. I would consider purchasing a product with a money back offer - but only if the price before rebate was the best, after considerations of convenience and risk. My automatic assumption is that I will not receive the rebate - either due to a fraudulent manipulation of the terms of sale or because the vendor expects to file bankruptcy. If the offer was honest the whole transaction would be completed at time of sale - the only reason to introduce delay is to defraud me of the balance payment I might otherwise have expected.
Another pet-hate of mine is supermarket loyalty cards. When one is pushed in my direction I normally state (loudly) that I am not loyal to supermarkets! I consider convenience quality and price only - and if the supermarket wishes to complicate this issue by making me jump through hoops to receive a fair deal then I am far more likely move my business elsewhere. If they want _my_ loyalty they should stock quality products with clear and honest prices and I will return - annoy me with unnecessary penny-pinching nonsense and I can only loose respect for the intentions of the retailer. I realise that my comments will unlikely make a difference - but I consider it my selfish good deed. In the news last year a retired man discovered that he could make profit buying bananas because they earned a substantial number of loyalty points. The story went that he decided to use this to pay for his shopping, and over the period of a couple of months he calculated that all his shopping had been free, only requiring that he dispose of around 20 tonnes of bananas. The industrial banana issue would have put others off - but not this tenacious pensioner! He hand-delivered them to everyone and anyone he could find - hence in a single amusing move taking a swipe at deceptive marketing and making friends into the bargain. My hat was off to him - my hero!
We disagree where you suggested that customers who shunned small retailers introduced the dif
I was about to point out in response to "A big advantage of statistics-based interfaces is that they automatically enforce correctness..." that rather than enforce correctness they will more likely introduce common errors.
When designing a language - be that a simple one which can be encapsulated in an XML schema for example, or even a complex natural language there is a trade off between being efficiently terse and introducing sufficient redundancy as to allow communicants to differentiate signal from noise. If you enter data in a format too terse then you are more likely to make errors which can't easily be detected - if you enter in a language too redundant you will find it tedious but errors are more likely detectable at a syntactic level.
For this reason I'd like to see exactly the opposite approach... I'd like to see long-hand ways of entering data where my errors are detected and flagged - which are then parsed and stored or transmitted in a more efficient format.
I can only speak for myself, but I for one think computing features are the best advancement in car technology in my lifetime. After many hours deliberating which nearly new car deserved my hard earned dosh, the Toyota choice won over Honda and Mitsubishi for one main reason - it had Sat-Nav built into the dashboard. I know I can buy PDA navigators, and DIN units to slot into the dash, but I don't want a device that's easy to steal when I park in less respectable neighbourhoods and I can't be bothered with plugging in electronic paraphernalia which would invariably have a flat battery when I unexpectedly need to use it. The Sat-Nav unit I have is one of the first generation - i.e. very basic. It has a small monochrome screen; few configurable parameters, no display of OS-like maps... The user interface plumbs new depths of dire - and you even have to enter addresses in long-hand because it doesn't understand postcodes. Conversely I wouldn't give up Sat-Nav. On the motorway (even for journeys I know quite well) I find the information about distance to next junction and real-time alerts for traffic jams with 're-route' extremely useful when deciding if I should change lanes - or to pull off the motorway and use the alternative 'pre-motorway' route most modern drivers have forgotten. While to be macho I need to pretend I can find my own way to places - I find that when Sat-Nav confirms my idea I feel more confident and concentrate better on my surroundings without worrying that I might overshoot my turn. When visiting new cities Sat-Nav comes into its own - calmly establishing a route from where you are now to your entered destination without the histrionics a human passenger-navigator normally brings to the experience - and making the idea of reading a street map while driving a nightmare firmly in the past. In-car computation is in its infancy - I want to be able to talk to my car - I want to give it instructions about my destination, I want to ask it questions - like "Please direct me to the nearest garage selling charcoal." or "Do I have time to go home first, stay for 10 minutes, then continue to arrive by 8pm?" The possibilities are endless... and if the technology and price is right my chequebook, for one, is open!
Hmmm - I'm not familiar with either of those (I presume) pre-processor definition symbols... I was thinking of the work relating KSE (Kernel Scheduled Entities) - that always seemed a very good idea to me - though I've not played with them. Care to expand on what you mean by SCHED_ULE and SCHED_4BSD?
Better TCP by whose rules?
on
Replacing TCP?
·
· Score: 3, Insightful
When considering protocols for information transport, it is very important to be absolutely sure what assumptions you are making. There are a number of non-independent factors which influence the suitability (and hence efficiency) of network protocols to application demands. Bandwidth, for example, is related to but doesn't define the statistical distribution of latencies; maximum packet rate and their relationship to packet size. The channel error rate (and statistical distributions of packet failures) are again linked to fragmentation and concatenation of transmitted datagrams - and this in turn affects latencies when considering "reliable" transport protocols. Routing policy and symmetry of physical links introduces yet more tradeoffs which need to be considered - not to mention the potential problems evaluating if the burden of protocol computations outweighs the advantage of an improved strategy for a given physical link. (And I'm not even going to mention security!) When considering protocols the most important thing to consider is the model they assume of the communications infrastructure on which they are to be deployed. TCP is likely the optimal solution given the assumptions TCP makes... if you change those assumptions to more closely fit a particular network infrastructure you will likely get better performance only on that infrastructure, but far worse performance where your new assumptions do not hold. I used to be interested in the idea of dynamically synthesizing protocols to best suit the actual physical links in a heterogeneous network... however my ideas were met with extreme disinterest; I felt my critics demanded I present a protocol which beats TCP under TCP's assumptions - and no amount of hand-waving and explanation would convince them this was a silly request. I still think the idea has merit - but having wasted 3 years of my life trying to push it uphill, I've found other interesting (and far more productive) avenues to pursue.
You're definitely onto a winning strategy there - well - sort of. It's almost a decade since my undergrad, when I was always one of the first to start assignments. The vast majority of the assignments followed a common trend. Predicted to take about 1-2 weeks to complete, and after spending 3 weeks+ the early starters would discover the problem demonstrably impossible... the most amusing incident involved modification of a natural language parser in Prolog - where it would take the departmental line printer 12 years for each student to print the documentary requirements for the first of 5 parts. This assignment was radically simplified and clarified - hence those who only started in the last week had a significant advantage over those who started early. I always decided I wanted to be the one to find the problems first (always suspecting that the occasion would arise where the assignment would be possible but too difficult to complete in the final few days - even given clarifications) but only once did that prove to be the case.
Judging by all the people I've met, the most important advice a computing student should take is to be VERY careful with any open-ended projects. In my final year everyone had to write an open-ended project... I chose mine to implement a compiler and runtime for a non-deterministic distributed language based upon Dijkstra's WP language which would survive network outages and fail-stop of individual processors. I worked around the clock for 9 months - I was stretching all the facilities at my disposal and while each problem I tackled was easily within my capabilities - the sheer number of time consuming steps was staggering. I attended the fewest lectures I could get away with (later relying upon revision time to learn much of the material required for my finals) and was to be found working on this project 7 days a week from 8am well into the small hours. With 48 hours to go before my formal presentation, the system as a whole would not demonstrate a credible execution (despite this having been my sole aim for over 6 months being painstakingly careful with the individual components.) After an all-nighter - a matter of hours to go before I had to "show all" I fixed the last bug preventing the demo... FAR too close for comfort. At the demo I couldn't help thinking the professors were more impressed with the coloured wool diagram on a mobile pin-board created just hours before than they were with the thousands of technical challenges I had faced down. If I were doing my final year again I'd pick a far more simple project - with no more than a month spent programming... this would have given me adequate time to make sure I understood the material which would feature in the courses assessed by examination and ample time to make the most of my project with slick presentation and substantially improved project report. While everything worked out great for me, I can't help thinking I did this the "hard way" - the most important advice I can give to any computer scientist who wrote code before starting their degree is that they must avoid being overly ambitious - the belief that you can tackle a large scale project to tight deadlines, when there are alternatives, is the biggest mistake you can make.
I believe FreeBSD 5.0 has a new scheduler - which particularly caught my eye - but also noteworthy is that BSD 5.0 has better hardware support than 4.x - specifically, for example it supports my 802.11g card allowing me to roll my own secure wireless gateway.
The FreeBSD folks would benefit from a clear document describing the differences between 4 and 5 - I'm sure they have one but it isn't presented anywhere prominently...
I stand corrected - though I maintain it is a bizarre example... Since no user-oriented tools exist to manage hard links on NTFS I suspect that they would be rare in practice. In any case - if such a hard-link did exist then I'd expect it would be fully considered in the design of the application making this bug appear somewhat less serious than it, in fact, happens to be.
Are you telling me that the executable search path is traversed when trying to determine which file temp.dat might be? That would seem a very silly thing to do - and not something I've noticed.
From the article : "c:\dir\test.dat, test.dat, and..\..\test.dat might all refer to the same file."
Now I could understand how c:\dir\test.dat and test.dat might be the same file - but, prey, assuming a hierarchical file system, how can all three be identical given that Windows file systems don't support hard-links? The test.dat is the same as the c:\dir\test.dat file if we are in c:\dir - but then the parent of the parent could not be c:\dir as well.
Slashdot Mirror
While it is no particular advantage to the open source movement to have open source tools on windows, it is of extreme benefit to have platform independent tools in use on Windows. Open source benefits substantially from the availability of non-proprietary software on the (proprietary) OS with the largest market share. The idea that Open Source advocates should be against open source programs on Windows is asinine. Is anyone really worried that Windows will transpire to be a superior OS to run platform independent software?
I'm not particularly enamoured with PDAs - particularly the ones which also try to be a phone. I have, however been trying to research my options for a device more portable than my laptop on which I can easily read and reply to email; use the web etc. It would need to have the handwriting recognition facilities of the best modern PDAs; have a substantial battery life - at least 8 hours; and finally, a c. 7" screen with at least SVGA resolution. I wouldn't mind if it was fairly heavy but I would find anything larger than c. A5 as cumbersome as a laptop when used on my lap in an sumptuous arm chair. I'm sure there's a market for such a device but I can't find one... I did find these - but while this is the format I want, this product is only a DVD player.
I'm 100% behind the idea that I should be able to filter threads with only one message... That would be very useful for me too.
I wonder - am I alone in seeing the "Google Groups 2" as a significant improvement on the original?
I like the improved 'posting' speed; I love the 'starred topics' (Though I remain sceptical that the 'new posts' feature works properly - I keep thinking "new since when?"). I like the idea that a thread has become the notional unit searched in the new UI - Google Groups 2 far better suits my needs.
In moments like these I'm glad I bought the tin foil hat and not the more luxurious titanium one.
Real cold fusion nut jobs would choose a palladium hat.
My own camera phone is on-order... so I can't say for sure... but I strongly suspect that the screen will prove inadequate for any complex text - I expect a phone number, street name or some such would be legible (from what I've seen on others' phones) - though for my own use I'm not so bothered - I can easily display the images on my desktop/laptop when I'm at home or in the office... which (luckily) happens to be when I find I wish I remembered such details.
I used to think exactly the same. I was given an SLR camera 20 years ago and I probably took a hundred or so photos with it as a teenager. 3 years ago I bought a 2MP point and shoot digital - and I've taken around 30 to 40 photos so far - and kept 2 or 3 (as digital images.) I'm no photographer - most of my digital photos were to document things - I bought the digital to document where someone had driven into my car then driven off. I thought "another camera - how could that possibly be valuable to me?"
Keeping that thought in mind - I've found dozens of occasions I'd have made use of one... just not for taking snaps of people, or places. I frequently find myself noticing a sign with a phone number or a closed shop or restaurant with opening hours posted in the window. I never seem to have a pen or paper (and copying down a list of opening hours would be very tedious anyway) - but a photo of the sign taken using my mobile would be perfect... it would take only a couple of seconds, and I wouldn't loose the information (unless I loose my phone.)
I expect to become a convert.
Plausible... but that doesn't explain why it is better that I have to blindly fumble for a working firmware where an OSS project could simply give me an MD5 to verify.
I can't argue with your pragmatism - and I guess I'd be far more inclined to take the same route if I started again now... like yourself I followed the handbook, then after that failed I tried other how-to documents... Ideally I'd like to see the handbook corrected so as not to misguide anyone else. I understand the concerns about performance - but to be honest performance isn't an issue for me... A reliable always-on cheap 28800 baud link would suit my needs...
In an attempt to explain my reasoning (which one might reasonably consider flawed) I decided to by a Speedtouch USB ADSL modem rather than an Ethernet one for several reasons:
So far I've discovered that the "Alcatel Speedtouch USB" modem isn't the same as a "Speedtouch 330 USB" modem - and, of the latter, the silver ones are "Rev 4" and need different firmware to that documented in most howto documents. I've not established why I can't establish a PPPoA connection using it.... Nor why the FreeBSD handbook seems to assume I'm using an Ethernet device (judging from the sample configuration) right under a heading specifying USB. I'd love to hear from someone who has a Speedtouch 330 Rev4 modem working using PPPoA either on Free/Open BSD... especially if they will share the specific details of their working configuration... all the documents I can find seem to be contradictory or assume that my USB modem transmogrifies into an IP addressable device.
It is only a for-home system... the Speedtouch modem works fantastically under windows - and hell - all I want it to do is chuck packets at my ISP from my BSD box. As the handbook suggests just such a configuration - I'm sure you'll agree it shouldn't be a big problem to set up.
I'd prefer not to use a "wireless router" as I want to use more advanced packet filtering and IPSEC encryptions which are easy with BSD. I like the conceptual simplicty of 1 ADSL adaptor; 1 Ethernet card; 1 wireless card. In order to achieve a comparable level of security but avoiding dedicated ADSL modems - I'd need a second ethernet card and an ADSL router (from which I'd use precious little functionality) as I really want my LAN insulated from the ADSL link by a gateway machine.
My budget isn't all that shoestring - but I'm loathed to spash out on new kit when I already own something that works perfectly well for my purposes. Obviously a commercial solution would be to buy in a working solution... but this isn't a commercial project.
I can't fault the preference for redistributable firmware - that's obvious. However, we don't need any third party to change its behaviour to overcome my hassles... just clear, accurate descriptions (with MD5 sums where using 3rd party components) of the components themselves and their configuration.
Downloading drivers is no hassle for me - and, I suspect, if the drivers worked painlessly and proved easily configured then the manufacturer would be happy to put the firmware on the CD distributed with the hardware. Redistributable would be best but I'd be happiest if we addressed the OSS system first in order that the only problem is redistribution of the firmware (which seems a relatively minor inconvenience right now.)
An amusing chicken and egg quandary for sure - but not a serious concern for most BSD/Linux users who usually have at least occasional access to another internet connected PC. I'd say misleading or missing documentation is a bigger issue - I certainly can't download firmware if I don't know which firmware I should be downloading!
Hmmm - that makes Atheros looks just as supported as in FreeBSD... My concerns (about a year ago) were that the only references to ATH in OpenBSD were messages suggesting that it wasn't supported - I didn't feel inclined to install it to check the manual pages.
I don't suppose you can also tell me that OpenBSD can also solve my problems with my "Speedtouch 330 (Rev 4)" USB ADSL modem assuming an ISP demanding PPPoA? If I can solve my headaches with a change from FreeBSD to OpenBSD, I'd do that in an instant.
Part of me wants to back Theo arguing for distributable firmware - but another part of me feels that there is still a lot that can be achieved without requiring any re-licensing.
I'm currently stuck trying to get my Alcatel/Thompson "Speedtouch 330" (Revision 4) ADSL modem to work under FreeBSD 5.3. Downloading the 'firmware' was a pain but much of that could have been resolved with some good documentation and an MD5 to verify the correct version. Even now I have the device recognised following the handbook doesn't get me connected... and offers precious little information about how to make appropriate configuration.
I suppose the response might be that that OpenBSD would do this fine - though I chose FreeBSD as a result about concerns about OpenBSD support for the Atheros chipset in my Dlink DWL G520 PCI wireless net card (which is straightforward to configure in FreeBSD.) Aaaagh!
I'm compelled to reply - your rant is off-base on at least a couple of points - though I concede that you've identified an interesting mentality. Here's my own rant!
Let me start by explaining how I value goods: I start by making sure I compare like-with-like, then arrive at a value of the item itself - this is the cheapest price at which I can buy the item. Next I factor in convenience (which, to me at least) is extremely important - and finally assess any risks (such as a risk of receiving faulty goods and any cost or effort involved in resolving the issue.) I compare my weighted prices and choose the cheapest (weighted) price - it is the "sane" way.
Things are more complicated when considering dissimilar products or (frequently) re-branded products at different prices. I find re-branding extremely counter productive... such a product may introduce compatibility concerns or be more difficult to repair or service - hence I assess risk adversely. Expanding the namespace is hardly convenient either. For example, I live in the UK, and 2 of the main electrical retailers are Currys and Comet, and it is normal to find both stores within easy walking distance of each other - they usually appear to be locked in a price war. Interestingly both stores are owned by the same group that owns Dixons (another large UK electrical retailer.) The two store names are, in my opinion, a scam to present a false impression of competition. I still buy from these stores but I have become extremely price and brand sensitive about their offers. When their sales-operatives enter their sales spiel for extended-warranty I reply with my own - asking if they are knowingly supplying defective goods; and ask if they intend to comply with UK consumer protection laws. The conversation usually ends with - if the goods turn out to be faulty after a year I will simply remember to replace with a product by a different manufacturer from a different retailer.
I _DESPISE_ money back offers - to me they scream "FRAUDULET VENDOR ALERT!" - so... I guess they are useful. I would consider purchasing a product with a money back offer - but only if the price before rebate was the best, after considerations of convenience and risk. My automatic assumption is that I will not receive the rebate - either due to a fraudulent manipulation of the terms of sale or because the vendor expects to file bankruptcy. If the offer was honest the whole transaction would be completed at time of sale - the only reason to introduce delay is to defraud me of the balance payment I might otherwise have expected.
Another pet-hate of mine is supermarket loyalty cards. When one is pushed in my direction I normally state (loudly) that I am not loyal to supermarkets! I consider convenience quality and price only - and if the supermarket wishes to complicate this issue by making me jump through hoops to receive a fair deal then I am far more likely move my business elsewhere. If they want _my_ loyalty they should stock quality products with clear and honest prices and I will return - annoy me with unnecessary penny-pinching nonsense and I can only loose respect for the intentions of the retailer. I realise that my comments will unlikely make a difference - but I consider it my selfish good deed. In the news last year a retired man discovered that he could make profit buying bananas because they earned a substantial number of loyalty points. The story went that he decided to use this to pay for his shopping, and over the period of a couple of months he calculated that all his shopping had been free, only requiring that he dispose of around 20 tonnes of bananas. The industrial banana issue would have put others off - but not this tenacious pensioner! He hand-delivered them to everyone and anyone he could find - hence in a single amusing move taking a swipe at deceptive marketing and making friends into the bargain. My hat was off to him - my hero!
We disagree where you suggested that customers who shunned small retailers introduced the dif
I was about to point out in response to "A big advantage of statistics-based interfaces is that they automatically enforce correctness..." that rather than enforce correctness they will more likely introduce common errors.
When designing a language - be that a simple one which can be encapsulated in an XML schema for example, or even a complex natural language there is a trade off between being efficiently terse and introducing sufficient redundancy as to allow communicants to differentiate signal from noise. If you enter data in a format too terse then you are more likely to make errors which can't easily be detected - if you enter in a language too redundant you will find it tedious but errors are more likely detectable at a syntactic level.
For this reason I'd like to see exactly the opposite approach... I'd like to see long-hand ways of entering data where my errors are detected and flagged - which are then parsed and stored or transmitted in a more efficient format.
I can only speak for myself, but I for one think computing features are the best advancement in car technology in my lifetime. After many hours deliberating which nearly new car deserved my hard earned dosh, the Toyota choice won over Honda and Mitsubishi for one main reason - it had Sat-Nav built into the dashboard. I know I can buy PDA navigators, and DIN units to slot into the dash, but I don't want a device that's easy to steal when I park in less respectable neighbourhoods and I can't be bothered with plugging in electronic paraphernalia which would invariably have a flat battery when I unexpectedly need to use it. The Sat-Nav unit I have is one of the first generation - i.e. very basic. It has a small monochrome screen; few configurable parameters, no display of OS-like maps... The user interface plumbs new depths of dire - and you even have to enter addresses in long-hand because it doesn't understand postcodes. Conversely I wouldn't give up Sat-Nav. On the motorway (even for journeys I know quite well) I find the information about distance to next junction and real-time alerts for traffic jams with 're-route' extremely useful when deciding if I should change lanes - or to pull off the motorway and use the alternative 'pre-motorway' route most modern drivers have forgotten. While to be macho I need to pretend I can find my own way to places - I find that when Sat-Nav confirms my idea I feel more confident and concentrate better on my surroundings without worrying that I might overshoot my turn. When visiting new cities Sat-Nav comes into its own - calmly establishing a route from where you are now to your entered destination without the histrionics a human passenger-navigator normally brings to the experience - and making the idea of reading a street map while driving a nightmare firmly in the past. In-car computation is in its infancy - I want to be able to talk to my car - I want to give it instructions about my destination, I want to ask it questions - like "Please direct me to the nearest garage selling charcoal." or "Do I have time to go home first, stay for 10 minutes, then continue to arrive by 8pm?" The possibilities are endless... and if the technology and price is right my chequebook, for one, is open!
Hmmm - I'm not familiar with either of those (I presume) pre-processor definition symbols... I was thinking of the work relating KSE (Kernel Scheduled Entities) - that always seemed a very good idea to me - though I've not played with them. Care to expand on what you mean by SCHED_ULE and SCHED_4BSD?
When considering protocols for information transport, it is very important to be absolutely sure what assumptions you are making. There are a number of non-independent factors which influence the suitability (and hence efficiency) of network protocols to application demands. Bandwidth, for example, is related to but doesn't define the statistical distribution of latencies; maximum packet rate and their relationship to packet size. The channel error rate (and statistical distributions of packet failures) are again linked to fragmentation and concatenation of transmitted datagrams - and this in turn affects latencies when considering "reliable" transport protocols. Routing policy and symmetry of physical links introduces yet more tradeoffs which need to be considered - not to mention the potential problems evaluating if the burden of protocol computations outweighs the advantage of an improved strategy for a given physical link. (And I'm not even going to mention security!) When considering protocols the most important thing to consider is the model they assume of the communications infrastructure on which they are to be deployed. TCP is likely the optimal solution given the assumptions TCP makes... if you change those assumptions to more closely fit a particular network infrastructure you will likely get better performance only on that infrastructure, but far worse performance where your new assumptions do not hold. I used to be interested in the idea of dynamically synthesizing protocols to best suit the actual physical links in a heterogeneous network... however my ideas were met with extreme disinterest; I felt my critics demanded I present a protocol which beats TCP under TCP's assumptions - and no amount of hand-waving and explanation would convince them this was a silly request. I still think the idea has merit - but having wasted 3 years of my life trying to push it uphill, I've found other interesting (and far more productive) avenues to pursue.
You're definitely onto a winning strategy there - well - sort of. It's almost a decade since my undergrad, when I was always one of the first to start assignments. The vast majority of the assignments followed a common trend. Predicted to take about 1-2 weeks to complete, and after spending 3 weeks+ the early starters would discover the problem demonstrably impossible... the most amusing incident involved modification of a natural language parser in Prolog - where it would take the departmental line printer 12 years for each student to print the documentary requirements for the first of 5 parts. This assignment was radically simplified and clarified - hence those who only started in the last week had a significant advantage over those who started early. I always decided I wanted to be the one to find the problems first (always suspecting that the occasion would arise where the assignment would be possible but too difficult to complete in the final few days - even given clarifications) but only once did that prove to be the case.
Judging by all the people I've met, the most important advice a computing student should take is to be VERY careful with any open-ended projects. In my final year everyone had to write an open-ended project... I chose mine to implement a compiler and runtime for a non-deterministic distributed language based upon Dijkstra's WP language which would survive network outages and fail-stop of individual processors. I worked around the clock for 9 months - I was stretching all the facilities at my disposal and while each problem I tackled was easily within my capabilities - the sheer number of time consuming steps was staggering. I attended the fewest lectures I could get away with (later relying upon revision time to learn much of the material required for my finals) and was to be found working on this project 7 days a week from 8am well into the small hours. With 48 hours to go before my formal presentation, the system as a whole would not demonstrate a credible execution (despite this having been my sole aim for over 6 months being painstakingly careful with the individual components.) After an all-nighter - a matter of hours to go before I had to "show all" I fixed the last bug preventing the demo... FAR too close for comfort. At the demo I couldn't help thinking the professors were more impressed with the coloured wool diagram on a mobile pin-board created just hours before than they were with the thousands of technical challenges I had faced down. If I were doing my final year again I'd pick a far more simple project - with no more than a month spent programming... this would have given me adequate time to make sure I understood the material which would feature in the courses assessed by examination and ample time to make the most of my project with slick presentation and substantially improved project report. While everything worked out great for me, I can't help thinking I did this the "hard way" - the most important advice I can give to any computer scientist who wrote code before starting their degree is that they must avoid being overly ambitious - the belief that you can tackle a large scale project to tight deadlines, when there are alternatives, is the biggest mistake you can make.
I believe FreeBSD 5.0 has a new scheduler - which particularly caught my eye - but also noteworthy is that BSD 5.0 has better hardware support than 4.x - specifically, for example it supports my 802.11g card allowing me to roll my own secure wireless gateway.
The FreeBSD folks would benefit from a clear document describing the differences between 4 and 5 - I'm sure they have one but it isn't presented anywhere prominently...
I stand corrected - though I maintain it is a bizarre example... Since no user-oriented tools exist to manage hard links on NTFS I suspect that they would be rare in practice. In any case - if such a hard-link did exist then I'd expect it would be fully considered in the design of the application making this bug appear somewhat less serious than it, in fact, happens to be.
Are you telling me that the executable search path is traversed when trying to determine which file temp.dat might be? That would seem a very silly thing to do - and not something I've noticed.
From the article : "c:\dir\test.dat, test.dat, and ..\..\test.dat might all refer to the same file."
Now I could understand how c:\dir\test.dat and test.dat might be the same file - but, prey, assuming a hierarchical file system, how can all three be identical given that Windows file systems don't support hard-links? The test.dat is the same as the c:\dir\test.dat file if we are in c:\dir - but then the parent of the parent could not be c:\dir as well.