Well, all I can say is I'm happy I don't live in the only completely backwards developed country in the world anymore! Welcome to the bottom rung of the ladder UK! I think there's room here next to us lowly US techies.
the examples in the article are general timer based dtrace calls, the one specifically that he was using was seeing the most accessed files every 10 seconds. Well, if iTunes is running on the CPU at the 10 second interval, then the probe fails to run, and you don't get the data you were trying to get, even though it has nothing at all to do with iTunes.
Anything you do with dtrace which is designed to gather stats like this would be effected by this bug, your stats will be all out of whack, because basically the apple dtrace says "if itunes is running on the CPU, don't dtrace", it doesn't say "if they are trying to dtrace itunes, disallow it"
I agree with your general assertion that the market is broad, and knowing any given specific technology may not be the best for whichever job you eventually land in.
However, having worked with both types of programmers (IE those who went through a rigorous CS program and those who went through a "here's java it'll solve all your problems" program) regardless of what language the current project is in, the rigorous CS program types are better hands down, every time.
Yeah you may not know exactly how the JRE is allocating memory, but you are able to look at a chunk of code and say "well holy geeze, this program is going to have to allocate a whole ton of memory right there where its loading a select * query directly into a single array... and lets see yeah that table has 4.8million rows in it... " This sort of thinking completely baffles and befuddles most "java only" type programmers I've met. They consistently and repeatedly make these kinds of mistakes because they haven't ever been exposed to lower level programming. They don't even grasp the concept that RAM is finite, that CPU cycles are finite (well, I just ask the computer for more RAM, and it gives it to me..).
Most of my projects these days are in python, yeah yeah its high level, and its interpreted, so its slow! Well, sure, except you can escape out to C if what you are trying to do really needs the speed. It has garbage collection so you don't *HAVE* to think about memory allocation. But I always do. I always run my code through profilers, test it on a large load, make sure it is acting sanely both in CPU time and memory consumption.
This is not what Java or these dumbed down CS curricula teach students. Besides the fact that just firing up the JRE eats 256-512MB of ram, so you wouldn't even notice if your app which is adding up invoice totals and should only need 512k of RAM to do its work is actually using 5-20MB of RAM because you're loading whole tables from the DB into memory, which isn't a problem on the test DB (only 5-20MB), but when you throw that program at the production DB with 2.8TB of data... well most computers throw up when you try to load that into RAM.
My point is simply that learning bigO, memory allocation, and the general trade offs between CPU usage and Memory usage which used to be taught in any decent CS program these things are 100% necessary to any programming career. Also, data structures and basic algorithms are good too, even if you aren't going to write a system to generate btrees, its good to know how a database index works... It's good to understand why regenerating indexes takes a long time, its good to know what operations take a lot of memory or a lot of CPU and WHY they take them.
I learned programming in Pascal, C, and C++, I've never touched assembly. I feel I am an effective Java, c#, and python programmer BECAUSE I understand the basics of what the underlying system is doing. So sure in a higher level language you don't know the exact implementation of how it is dealing with dynamically allocated RAM, but you know its happening, and you understand the basic concepts of linked lists, heaps, stacks, and pointers and how memory management is probably being done. Without these basic foundations I don't think you can be a very effective programmer in any language.
The answer is phone services have a huge barrier to entry. You need somewhere in the neighborhood of 5-10 billion dollars to provide service (IE infrastructure, fiber, routers, etc) to all the homes in a single medium sized state. Multiply that by 50, and you need 250-500 billion dollars in capital to do the whole country.
Then, you have to convince/bribe/cajoul the politicians in each state/city to give you rights of way so you can lay your fiber. This is going to cost another 5-10 billion dollars country wide. Then you have to spend another 50-100 billion in marketing to steal customers away from the incumbents. Now, maybe, if you're lucky and your marketing worked, and the incubents didn't drop their prices by 50%, you have a revenue stream.
Once you have the infrastructure in place, adding MB of bandwidth is relatively cheap. This is why everyone hates the phone/cable companies, because they have no competition, but they could be providing much better services for the same we are paying now, but they dont' have to because there is not competition. If someone could magically lay a fiber network for cheap, they could destroy the incumbents. But the incumbents know this as well, and if someone did try to compete, they could just as easily drop their prices and keep customers from switching, causing the new provider to fail. They would then buy up the new providers assets at a bankruptcy fire sale, and then raise prices back up and not use the new assets to provide more services.
As someone who has inherited quite a few code bases in the state you describe in your last paragraph (and trades in turning around large projects which have gone off the tracks), I can completely agree with you. If there isn't a decent design behind the system, something that can be explained in a few days or a week, that details the major modules of the system, the major code/data paths in the system, and the overall design philosophy, then it gets very difficult.
In general projects in that state have/had 1 or 2 people who are holding the project/company for ransom (or they are just idiots). They've built the system with the thought in mind that no one can figure this out, so we have jobs for life (or with no thoughts in their minds). Either someone comes in and fixes it, provides a design element to the project or the project dies.
Also, I agree with your overall sentiment, you have to learn the design/big picture first. Individual code details give you nothing if you don't know the big picture. In fact, knowing code details without the big picture can be very harmful.
I feel that Rails did make a large contribution to web development in general. There are now similar frameworks in basically every language out there. Even Java has a RoR look a like (JBoss Seam).
However, the biggest issue I feel is the whole "Convention over configuration" mantra. Yes a framework should have sane defaults (convention), however, in my experience using Rails, Django, Seam, and PHPCake, Rails by far makes it harder to do something that is not a "default". Want to use it with an existing DB? django and seam both have a beautiful utility that will build your code from that DB. RoR says: you must rename all your db tables to plural (which I was taught in school was 100% wrong btw... I was always taught singular for table names... but maybe my professors were smoking something? At any rate, all of my existing DBs had singular table names when I started messing with Rails), or spend hours adding configuration to your app to properly map models to tables.
There are many other issues I ran into using Rails along these lines and others, but basically it always seemed like in Rails it is very much an all or nothing proposition. Either your application/design/db fits and it is all going to work, or there is need xyz which is outside of Rails features, and therefore, you can't use any of rails.
Django is my favorite of the 4 frameworks listed, specifically because it makes it so easy to use parts and pieces of the framework, where it makes sense, or not use it for other parts. It is by far the most flexible to me. And yes, there is some configuration needed, but not much more than RoR. The nice thing is django has a flexible and powerful configuration system. it isn't a second class citizen like most configuration in RoR (if a configuration option is even provided in RoR). Most of the time in RoR it seems like they said "Well, this is HOW you do this" and if you want to do it a different way, there isn't even an option anywhere to change it. You can either re-write the RoR functionality to match how you want to do it, or not use RoR. In reality though, these changes could be implemented as a simple configuration change. But because RoR eschews configuration like the plague, you're just stuck with 2 bad choices (reimplement, or don't use RoR).
This isn't to say RoR sucks, I'm just pointing out my reasoning for using a competing product.
This is my number 1 complaint against Rails. I used it for a couple side projects, couldn't get anything done (yes I got it running, I wrote a bunch of code, but as soon as I got into the real meat of the projects, it just turned ugly).
I moved to django, and all the walls were gone, I could concentrate on solving problems instead of staying within the lines that Rails draws.
DHH's mantra of "convention over configuration" is what creates this problem I think. A lot of times a well designed configuration system can be much more productive, expressive, and flexible than even the best intentioned convention. Simply put, a convention decided by 1 or a few developers is not going to be the right choice for 100% of projects or systems. It probably isn't even the right choice for 50% of projects. However the Rails guys have the ego and audacity to declare "This is the one right way!".
I agree with the parent post, in my Rails days, every headache I ran into had to do with hard coded, poorly designed "features" in ActiveRecord.
I've since moved on to Django, and yes, the framework doesn't do *everything* for you. But it is specifically designed not to. The project has made decisions about quite a few things along these lines "Well, that's just too hard to get right for everyone, so its not in the framework here's how you do this directly, don't use the framework for x".
In my opinion this is Rails greatest shortcoming. They are trying to make the framework do everything, and in so doing, they take some pretty serious shortcuts that then have the effect of mandating that you make poor design decisions or have a bunch of work to do if you need to do something *differently* than they have decided is the "one true way".
Yeah... That is a great lie too. Ever tried actually uploading a large file? Comcast has a 1mbps "burst" for the first 45 seconds to 1 minute (enough to fool all the speed test utilities). It then throttles back to 256kbps, so if you are actually moving a large file, the first 4-5MB go pretty fast, then the brakes get turned on, and your 20MB file takes 2 hours to upload.
This is not a "light-driven" reaction. It is a heat driven reaction. This beer keg sized device has 88m2 of mirrors concentrating solar radiation into it to heat cobalt above 2800 degrees.
There is a heat variation which could be used to heat water to drive a turbine in concert with this. The cobalt is heated to greater than 2800 degrees, it then must be cooled below 2000 degrees. This process must be repeated over and over again to continually capture CO2.
You have to remember this device is the size of a beer keg (IE not very big) and they specifically mention using many of these devices at a single coal fired plant.
Each device can grab 45lbs of CO2 and produce 2.5 gallons of fuel a day. Either you could increase the size of the device or increase the number of them to scale this. If you had 1000 of these devices at a coal plant, you would grab 45k lbs of CO2 a day and produce 2500 gallons of fuel a day. Still not a ton of fuel, but, its something. The reaction itself seems extremely simple, heat the cobalt til it releases some oxygen, cool and add CO2, the cobalt now reclaims oxygen from the CO2, leaving CO. The only way to "increase" this reaction as I see it would be to increase the amount of cobalt so that it has a greater potential to reclaim oxygen.
Why not use waste heat from the coal plant to run the CR5s? It seems to me that the coal plant itself has to be dumping many many BTUs out as waste heat and we should harness that to recapture and convert the CO2.
Although, I like your idea better being a general supporter of nuclear power, and seeing as how we could use a breeder reactor for this purpose, we could kill 3 stones with one plant.
1) clean CO2 out of the air and convert it to gasoline 2) generate electricity (burning coal and using the nuclear reactor's waste heat) 3) consume our existing nuclear waste
Do these batteries charge instantly?! I don't assume that there is no downtime between cars, I just don't think it takes anywhere near 10 minutes to pull your car up, pull out your CC swipe it and stick the pump into your car. You are making a big deal of the time between charges. If the station can charge their batteries in the time it takes for 1 car to replace another at the charging station then you don't need storage at all, it doesn't take 30 minutes to change cars, it might take 2-3 minutes. If you can replace the electricity in the batteries for a car charge in 2-3 minutes then why do you need the storage at all? It takes 10 minutes to deliver the charge, if you can grab that charge off the grid in 2 minutes, obviously you don't need storage. Even if your high estimate of 10 minutes to change cars at the pump is accurate... you still don't need storage if you can replace the charge in the batteries that fast. However, if it takes more like 60 minutes to replace the charge in the batteries, which I bet is more accurate, then having enough battery for 1 hour of charging is completely insufficient. if you have 2-3 hours of heavy usage, you will run out of storage, and your "quick charge" station will be relegated to a slow home charge station.
Also, you mention 40kWh each as the draw at the pump. We have already established that you need more than 300kW draw to charge the tesla in 10 minutes (600kW draw to charge a 400 mile range car in 10 minutes). Figure in an hour 4 cars can charge at each station (40 minutes of charging, 20 minutes for paying and moving cars), that is 2/3h*300kW = 200kWh per pump (400kWh for a 400 mile car) 5-10 times your estimate of 40kWh. So your assumptions need to be increased by 5-10 times. You mention 2mWh of storage, if you have 2mWh of storage and each pump is delivering 200kWh well, then you can run 10 pumps for an hour, you can provide a total of 40 charges per hour, for the first hour. At this point, your storage is nearly depleted, you have only the charge you were able to receive from the grid in the last hour... I don't know what a regular connection the grid can provide, but I'm sure it is no where near 2mWh. When do the batteries in your scenario get charged? If you have a run and everyone is trying to charge their cars, when is the downtime to charge these 2mWh of battery? Unless you can replenish the charge in those batteries in the time it takes to swap cars your storage will be depleted quickly. If you can charge those batteries in the time it takes to swap cars, then you don't need batteries at all because obviously you can draw the necessary power from the grid in near real time. However, it is a fact that the current grid cannot support this level of distribution. So I would argue that in reality for 16 charging stations you would probably need at least 5 times your peak draw in storage, 200kWh/pump * 16 pumps = 3.2mWh, so if you have 3.2mWh of battery then you can charge 4 cars/pump for a whole hour or 64 cars in an hour. Unfortunately, after charging those 64 cars, your storage is depleted, and assuming a regular electrical grid connection, it will take a number of hours to replenish the storage. For those hours you are effectively out of business, maybe you can charge 1 car at a time, but you will lose lots of business as you cannot charge a car at each pump because you are relying on this set of batteries to build up a charge to be able to deliver a high level of power quickly (in essence they are a large capacitor). On peak driving days the pumps can run all out for 4-5 hours straight (I'm not saying this is every day, but I know for a fact from experience that at least in the area I live, fridays before three day weekends, and days before holidays it does happen regularly). To support 5 hours of continuous 64 car/hr charging you would need 5*3.2mWh of battery, that is 16mWh or 8 times what you specced out, costing $640,000. Then hopefully you have some downtime where the draw does not exceed your charging capacity (the amount of powe
I specifically stated a "peak" draw of 5MW. No I don't think all pumps will be pumping at the same time 95% of the time, but the station has to be able to support that draw at the peak.
If the cells charge in parallel, then obviously the draw will have to be much higher for suv/minivan.
Sure, the station could have a storage system, so could a house. I've built data centers. I know how much battery arrays cost, and you have to replace them every 2-3 years. There is a reason why data centers have only enough battery to cover themselves for the couple minutes it takes to start a generator. Batteries are extremely expensive and have a huge recurring cost associated with them.
Sure most of the time you don't need the peak power, but there are plenty of times I've seen big truck stop gas stations with 16-30 pumps where every pump is running and cars are lined up 2-3 deep at every pump. If you are relying on a storage system to deliver that power you are going to run out of electricity, and then you are out of business. Sure you don't need this amount of power for every day, but on the big driving weekends when EVERYONE is getting in their car, well, your storage capacity of 3 times the needed peak load is not going to help you much. After the 3rd set of cars, you're out of power. To really support the sorts of peak loads I've seen you would need probably 10-15 times the batteries needed for a single set of charges across your pumps. This amount of batteries is certainly cost prohibitive. Friday afternoon before Memorial day weekend, Wednesday afternoon before Thanksgiving, these are the peak loads I'm talking about.
again I'm not against EVs, I would love to have one for city driving. But I do think there are substantial obstacles to wide spread adoption (IE >50% of cars are EVs and they are being used as primary family cars for road trips, and everything else, not just commuting), and I don't think the solutions are cheap, easy, or going to be readily available in the next decade.
That plan calls for covering 19% of the southwest with solar installations. I would say that is quite a bit larger than 22% of california.
And it only provides 69% of the electricity (oddly, only 35% of the energy... so we still need 65% fossil fuels?) I thought the idea of all this was to get off of fossil fuels by getting all cars, homes, businesses using clean electricity for heating, cooling and transportation.
So, we'd need to cover 25% of the southwest (not just california) for 100% of current electricity. Unfortunately they don't take into account the instant 25-40% increase in electricity usage that will occur when we all plug in our EVs for the first time.
Also, it is talking about PV. solar2 and solar tres are not PV installs, they are reflected light focused on a central tower. So, my question still stands, does the solar tech in the article not scale?
sorry to reply again, but thought of something else.
Even if lithium batteries increase their capacity by 10 times, there is still the problem of the life of the battery. Toyota recommends that you change the battery in the prius every 2 years, at a cost of $4500. Well, that $4500 dwarfs my gasoline costs for 2 years even without the power cost. This hidden cost is rarely if ever discussed when talking about EV or hybrid vehicles, but it is there. It is the reason I didn't buy a prius last year when I was shopping, the dealership was very up front and open about this cost, and stated that after 2 years the battery is almost useless and that the car would run almost 100% on gas unless you replace it. This cost makes any EV or hybrid vehicle completely uneconomical.
New lion batteries have not addressed the aging issue. My 12 month old laptop has seen about a 70% decrease in battery life. When cars are experiencing this phenomena people will not put up with it, and the batteries packs in cars will cost 8-10k once they are big enough to provide 400 mile range.
While I would love an EV and I think my driving patterns would be perfect for it (about 30 miles a day, no problem plugging the car in and letting it charge for the night), I would still have to keep a gas car for road trips (I take 1-2 1000 mile road trips for work per quarter).
Further, you don't address the problem the other poster brought up. 240V 1325Amp circuits are not going to happen, they are especially NOT going to be installed in anyone's home (instant death to anyone who accidentally touches that outlet). That is what you would need for a 10 minute charge of the tesla which only has a range of 221 miles. You HAVE TO HAVE at least 1 240V 1325Amp circuit and it simply isn't going to happen. If you double that range to 440 to make it competitive with gasoline powered cars, sure the battery tech might be here soon to support that, but you still need 2 times the energy. So now you need a 240V 2650Amp circuit to get a 10 minute charge?
As for "gas station" charging stations, figure 16 "pumps", 318kw * 16 pumps = 5MW of peak power. No gas station on the planet has a line running into it that can supply 5MW of power. Upgrading the electrical grid to support this kind of distribution will cost many billions of dollars. It is like an upgrade from copper to fiber in telecom. You need new transformers everywhere, new lines, new termination equipment at the customer prem, new equipment at the customer prem to be able to supply 16 circuits capable of carrying 318kw without melting. Sure it might be cheaper than upgrading the oil distribution infrastructure... but not by a huge percentage. It isn't going to be "cheap". Again multiply everything by 2 once the batteries are here which will actually support comparable range. So now you need a 10MW line running into every gas station to support 16 simultaneous 10 minute charges.
Also, I'll nitpick here a bit, it takes me about 3 minutes to gas up my 10 gallon car (the car gets 35 miles/gallon so 350 mile range). Taking 3 times as long to fill up is probably ok for a little commuter car, what about a family car like a minivan or small suv? They will obviously need a lot more power. My parent's SUV takes ~12 minutes to fill with gas, 3 times as long is 36 minutes. I don't know anyone who will stop a family vacation with kids in the car for 40 minutes to fill up with gas.
Also, what do you do about "running out of gas" in an EV? This will happen, it happens with gas, it will happen with electrical. With gas you go get a can and fill it up. Do you go buy a spare 200lb battery pack? Do you cart around a gas generator in the trunk? They will have to invent some sort of "quick charge" device that weighs less than 10lbs, can be purchased for $3 and provides enough juice for 20-30 miles (IE a gallon gas can replacement).
Umm... Not to burst your bubble, but that calculator says that it would take 34 years to pay back a 3KW system in Las Vegas, NV..
Unfortunately the cells only last for 25 years. That is not breaking even, because 9 years before you realize a return on your investment, you have to make the investment again. If you can't break even in Las Vegas, you can't break even anywhere in the US... well maybe death valley?
writing off a factor of 3 or 4 is still bizarre. I agree the original projection of 30 times was exaggerated. However, my electrical bill is ~$100/mo, you freely admit that if we moved to solar my bill would go to $300-400/mo... That is 100% unacceptable.
Just read the wikipedia on solar2, it said they are building solar tres in spain based on the design of solar2.
It says solar 2 is a 10MW plant, and that for solar tres they are DOUBLING the size (in area of mirrors) yet it will only be a 15MW plant... So... does this tech completely not scale? A doubling in size only creates a 50% increase in production? Double again and you only get a 22.5MW plant and now you're using 4 times the area of a 10MW plant?
I'm just saying, I doubt your straight linear math works, I bet it is a whole lot more than 22% of Cali to power the US because apparently this doesn't scale.
From a later post, I guess Windows Home Server *automatically* backs up clients that are connected to it, so you have your pretty little PC you open up your pictures and crop them... and whatever, Home Server at some point will back that up (corrupting the files in the process).
Google Reader begins sharing public data in a new way.
These were not "private" feeds, they were publicly available URLs (although obfuscated).
I'm not necessarily siding with Google on this one. I do think they should have thought this change of functionality out a little more, but the fact remains this data was already public. Comparing it to the Beacon scandal is not accurate at all.
yeah I'm retarded, $5k/kW sorry for the really poor math. The quotes I've received are ~40k including installation and it would still be probably $10k-15k after installation for these new panels too.. installation, mounting, labor is ~half of any solar install from what I've seen.
Slashdot Mirror
Well, all I can say is I'm happy I don't live in the only completely backwards developed country in the world anymore! Welcome to the bottom rung of the ladder UK! I think there's room here next to us lowly US techies.
the examples in the article are general timer based dtrace calls, the one specifically that he was using was seeing the most accessed files every 10 seconds. Well, if iTunes is running on the CPU at the 10 second interval, then the probe fails to run, and you don't get the data you were trying to get, even though it has nothing at all to do with iTunes.
Anything you do with dtrace which is designed to gather stats like this would be effected by this bug, your stats will be all out of whack, because basically the apple dtrace says "if itunes is running on the CPU, don't dtrace", it doesn't say "if they are trying to dtrace itunes, disallow it"
I agree with your general assertion that the market is broad, and knowing any given specific technology may not be the best for whichever job you eventually land in.
However, having worked with both types of programmers (IE those who went through a rigorous CS program and those who went through a "here's java it'll solve all your problems" program) regardless of what language the current project is in, the rigorous CS program types are better hands down, every time.
Yeah you may not know exactly how the JRE is allocating memory, but you are able to look at a chunk of code and say "well holy geeze, this program is going to have to allocate a whole ton of memory right there where its loading a select * query directly into a single array... and lets see yeah that table has 4.8million rows in it... " This sort of thinking completely baffles and befuddles most "java only" type programmers I've met. They consistently and repeatedly make these kinds of mistakes because they haven't ever been exposed to lower level programming. They don't even grasp the concept that RAM is finite, that CPU cycles are finite (well, I just ask the computer for more RAM, and it gives it to me..).
Most of my projects these days are in python, yeah yeah its high level, and its interpreted, so its slow! Well, sure, except you can escape out to C if what you are trying to do really needs the speed. It has garbage collection so you don't *HAVE* to think about memory allocation. But I always do. I always run my code through profilers, test it on a large load, make sure it is acting sanely both in CPU time and memory consumption.
This is not what Java or these dumbed down CS curricula teach students. Besides the fact that just firing up the JRE eats 256-512MB of ram, so you wouldn't even notice if your app which is adding up invoice totals and should only need 512k of RAM to do its work is actually using 5-20MB of RAM because you're loading whole tables from the DB into memory, which isn't a problem on the test DB (only 5-20MB), but when you throw that program at the production DB with 2.8TB of data... well most computers throw up when you try to load that into RAM.
My point is simply that learning bigO, memory allocation, and the general trade offs between CPU usage and Memory usage which used to be taught in any decent CS program these things are 100% necessary to any programming career. Also, data structures and basic algorithms are good too, even if you aren't going to write a system to generate btrees, its good to know how a database index works... It's good to understand why regenerating indexes takes a long time, its good to know what operations take a lot of memory or a lot of CPU and WHY they take them.
I learned programming in Pascal, C, and C++, I've never touched assembly. I feel I am an effective Java, c#, and python programmer BECAUSE I understand the basics of what the underlying system is doing. So sure in a higher level language you don't know the exact implementation of how it is dealing with dynamically allocated RAM, but you know its happening, and you understand the basic concepts of linked lists, heaps, stacks, and pointers and how memory management is probably being done. Without these basic foundations I don't think you can be a very effective programmer in any language.
The answer is phone services have a huge barrier to entry. You need somewhere in the neighborhood of 5-10 billion dollars to provide service (IE infrastructure, fiber, routers, etc) to all the homes in a single medium sized state. Multiply that by 50, and you need 250-500 billion dollars in capital to do the whole country.
Then, you have to convince/bribe/cajoul the politicians in each state/city to give you rights of way so you can lay your fiber. This is going to cost another 5-10 billion dollars country wide. Then you have to spend another 50-100 billion in marketing to steal customers away from the incumbents. Now, maybe, if you're lucky and your marketing worked, and the incubents didn't drop their prices by 50%, you have a revenue stream.
Once you have the infrastructure in place, adding MB of bandwidth is relatively cheap. This is why everyone hates the phone/cable companies, because they have no competition, but they could be providing much better services for the same we are paying now, but they dont' have to because there is not competition. If someone could magically lay a fiber network for cheap, they could destroy the incumbents. But the incumbents know this as well, and if someone did try to compete, they could just as easily drop their prices and keep customers from switching, causing the new provider to fail. They would then buy up the new providers assets at a bankruptcy fire sale, and then raise prices back up and not use the new assets to provide more services.
As someone who has inherited quite a few code bases in the state you describe in your last paragraph (and trades in turning around large projects which have gone off the tracks), I can completely agree with you. If there isn't a decent design behind the system, something that can be explained in a few days or a week, that details the major modules of the system, the major code/data paths in the system, and the overall design philosophy, then it gets very difficult.
In general projects in that state have/had 1 or 2 people who are holding the project/company for ransom (or they are just idiots). They've built the system with the thought in mind that no one can figure this out, so we have jobs for life (or with no thoughts in their minds). Either someone comes in and fixes it, provides a design element to the project or the project dies.
Also, I agree with your overall sentiment, you have to learn the design/big picture first. Individual code details give you nothing if you don't know the big picture. In fact, knowing code details without the big picture can be very harmful.
when your DB has 10 tables, sure... when it has 150 its a bunch of goofy code that you shouldn't have to write or maintain.
I feel that Rails did make a large contribution to web development in general. There are now similar frameworks in basically every language out there. Even Java has a RoR look a like (JBoss Seam).
However, the biggest issue I feel is the whole "Convention over configuration" mantra. Yes a framework should have sane defaults (convention), however, in my experience using Rails, Django, Seam, and PHPCake, Rails by far makes it harder to do something that is not a "default". Want to use it with an existing DB? django and seam both have a beautiful utility that will build your code from that DB. RoR says: you must rename all your db tables to plural (which I was taught in school was 100% wrong btw... I was always taught singular for table names... but maybe my professors were smoking something? At any rate, all of my existing DBs had singular table names when I started messing with Rails), or spend hours adding configuration to your app to properly map models to tables.
There are many other issues I ran into using Rails along these lines and others, but basically it always seemed like in Rails it is very much an all or nothing proposition. Either your application/design/db fits and it is all going to work, or there is need xyz which is outside of Rails features, and therefore, you can't use any of rails.
Django is my favorite of the 4 frameworks listed, specifically because it makes it so easy to use parts and pieces of the framework, where it makes sense, or not use it for other parts. It is by far the most flexible to me. And yes, there is some configuration needed, but not much more than RoR. The nice thing is django has a flexible and powerful configuration system. it isn't a second class citizen like most configuration in RoR (if a configuration option is even provided in RoR). Most of the time in RoR it seems like they said "Well, this is HOW you do this" and if you want to do it a different way, there isn't even an option anywhere to change it. You can either re-write the RoR functionality to match how you want to do it, or not use RoR. In reality though, these changes could be implemented as a simple configuration change. But because RoR eschews configuration like the plague, you're just stuck with 2 bad choices (reimplement, or don't use RoR).
This isn't to say RoR sucks, I'm just pointing out my reasoning for using a competing product.
This is my number 1 complaint against Rails. I used it for a couple side projects, couldn't get anything done (yes I got it running, I wrote a bunch of code, but as soon as I got into the real meat of the projects, it just turned ugly).
I moved to django, and all the walls were gone, I could concentrate on solving problems instead of staying within the lines that Rails draws.
DHH's mantra of "convention over configuration" is what creates this problem I think. A lot of times a well designed configuration system can be much more productive, expressive, and flexible than even the best intentioned convention. Simply put, a convention decided by 1 or a few developers is not going to be the right choice for 100% of projects or systems. It probably isn't even the right choice for 50% of projects. However the Rails guys have the ego and audacity to declare "This is the one right way!".
Ever worked with Django?
I agree with the parent post, in my Rails days, every headache I ran into had to do with hard coded, poorly designed "features" in ActiveRecord.
I've since moved on to Django, and yes, the framework doesn't do *everything* for you. But it is specifically designed not to. The project has made decisions about quite a few things along these lines "Well, that's just too hard to get right for everyone, so its not in the framework here's how you do this directly, don't use the framework for x".
In my opinion this is Rails greatest shortcoming. They are trying to make the framework do everything, and in so doing, they take some pretty serious shortcuts that then have the effect of mandating that you make poor design decisions or have a bunch of work to do if you need to do something *differently* than they have decided is the "one true way".
Yeah... That is a great lie too. Ever tried actually uploading a large file? Comcast has a 1mbps "burst" for the first 45 seconds to 1 minute (enough to fool all the speed test utilities). It then throttles back to 256kbps, so if you are actually moving a large file, the first 4-5MB go pretty fast, then the brakes get turned on, and your 20MB file takes 2 hours to upload.
This is not a "light-driven" reaction. It is a heat driven reaction. This beer keg sized device has 88m2 of mirrors concentrating solar radiation into it to heat cobalt above 2800 degrees.
There is a heat variation which could be used to heat water to drive a turbine in concert with this. The cobalt is heated to greater than 2800 degrees, it then must be cooled below 2000 degrees. This process must be repeated over and over again to continually capture CO2.
You have to remember this device is the size of a beer keg (IE not very big) and they specifically mention using many of these devices at a single coal fired plant.
Each device can grab 45lbs of CO2 and produce 2.5 gallons of fuel a day. Either you could increase the size of the device or increase the number of them to scale this. If you had 1000 of these devices at a coal plant, you would grab 45k lbs of CO2 a day and produce 2500 gallons of fuel a day. Still not a ton of fuel, but, its something. The reaction itself seems extremely simple, heat the cobalt til it releases some oxygen, cool and add CO2, the cobalt now reclaims oxygen from the CO2, leaving CO. The only way to "increase" this reaction as I see it would be to increase the amount of cobalt so that it has a greater potential to reclaim oxygen.
My first thought upon reading the article was:
Why not use waste heat from the coal plant to run the CR5s? It seems to me that the coal plant itself has to be dumping many many BTUs out as waste heat and we should harness that to recapture and convert the CO2.
Although, I like your idea better being a general supporter of nuclear power, and seeing as how we could use a breeder reactor for this purpose, we could kill 3 stones with one plant.
1) clean CO2 out of the air and convert it to gasoline
2) generate electricity (burning coal and using the nuclear reactor's waste heat)
3) consume our existing nuclear waste
Seems to be like a nearly perfect solution to me.
No, instead we make this coal/nuclear hybrid with a breeder reactor and it consumes our existing nuclear waste too!!!
It cleans the air of CO2, it cleans up nuclear waste, and it produces electricity and gasoline. Super Duper!
Do these batteries charge instantly?! I don't assume that there is no downtime between cars, I just don't think it takes anywhere near 10 minutes to pull your car up, pull out your CC swipe it and stick the pump into your car. You are making a big deal of the time between charges. If the station can charge their batteries in the time it takes for 1 car to replace another at the charging station then you don't need storage at all, it doesn't take 30 minutes to change cars, it might take 2-3 minutes. If you can replace the electricity in the batteries for a car charge in 2-3 minutes then why do you need the storage at all? It takes 10 minutes to deliver the charge, if you can grab that charge off the grid in 2 minutes, obviously you don't need storage. Even if your high estimate of 10 minutes to change cars at the pump is accurate... you still don't need storage if you can replace the charge in the batteries that fast. However, if it takes more like 60 minutes to replace the charge in the batteries, which I bet is more accurate, then having enough battery for 1 hour of charging is completely insufficient. if you have 2-3 hours of heavy usage, you will run out of storage, and your "quick charge" station will be relegated to a slow home charge station.
Also, you mention 40kWh each as the draw at the pump. We have already established that you need more than 300kW draw to charge the tesla in 10 minutes (600kW draw to charge a 400 mile range car in 10 minutes). Figure in an hour 4 cars can charge at each station (40 minutes of charging, 20 minutes for paying and moving cars), that is 2/3h*300kW = 200kWh per pump (400kWh for a 400 mile car) 5-10 times your estimate of 40kWh. So your assumptions need to be increased by 5-10 times. You mention 2mWh of storage, if you have 2mWh of storage and each pump is delivering 200kWh well, then you can run 10 pumps for an hour, you can provide a total of 40 charges per hour, for the first hour. At this point, your storage is nearly depleted, you have only the charge you were able to receive from the grid in the last hour... I don't know what a regular connection the grid can provide, but I'm sure it is no where near 2mWh. When do the batteries in your scenario get charged? If you have a run and everyone is trying to charge their cars, when is the downtime to charge these 2mWh of battery? Unless you can replenish the charge in those batteries in the time it takes to swap cars your storage will be depleted quickly. If you can charge those batteries in the time it takes to swap cars, then you don't need batteries at all because obviously you can draw the necessary power from the grid in near real time. However, it is a fact that the current grid cannot support this level of distribution. So I would argue that in reality for 16 charging stations you would probably need at least 5 times your peak draw in storage, 200kWh/pump * 16 pumps = 3.2mWh, so if you have 3.2mWh of battery then you can charge 4 cars/pump for a whole hour or 64 cars in an hour. Unfortunately, after charging those 64 cars, your storage is depleted, and assuming a regular electrical grid connection, it will take a number of hours to replenish the storage. For those hours you are effectively out of business, maybe you can charge 1 car at a time, but you will lose lots of business as you cannot charge a car at each pump because you are relying on this set of batteries to build up a charge to be able to deliver a high level of power quickly (in essence they are a large capacitor). On peak driving days the pumps can run all out for 4-5 hours straight (I'm not saying this is every day, but I know for a fact from experience that at least in the area I live, fridays before three day weekends, and days before holidays it does happen regularly). To support 5 hours of continuous 64 car/hr charging you would need 5*3.2mWh of battery, that is 16mWh or 8 times what you specced out, costing $640,000. Then hopefully you have some downtime where the draw does not exceed your charging capacity (the amount of powe
I specifically stated a "peak" draw of 5MW. No I don't think all pumps will be pumping at the same time 95% of the time, but the station has to be able to support that draw at the peak.
If the cells charge in parallel, then obviously the draw will have to be much higher for suv/minivan.
Sure, the station could have a storage system, so could a house. I've built data centers. I know how much battery arrays cost, and you have to replace them every 2-3 years. There is a reason why data centers have only enough battery to cover themselves for the couple minutes it takes to start a generator. Batteries are extremely expensive and have a huge recurring cost associated with them.
Sure most of the time you don't need the peak power, but there are plenty of times I've seen big truck stop gas stations with 16-30 pumps where every pump is running and cars are lined up 2-3 deep at every pump. If you are relying on a storage system to deliver that power you are going to run out of electricity, and then you are out of business. Sure you don't need this amount of power for every day, but on the big driving weekends when EVERYONE is getting in their car, well, your storage capacity of 3 times the needed peak load is not going to help you much. After the 3rd set of cars, you're out of power. To really support the sorts of peak loads I've seen you would need probably 10-15 times the batteries needed for a single set of charges across your pumps. This amount of batteries is certainly cost prohibitive. Friday afternoon before Memorial day weekend, Wednesday afternoon before Thanksgiving, these are the peak loads I'm talking about.
again I'm not against EVs, I would love to have one for city driving. But I do think there are substantial obstacles to wide spread adoption (IE >50% of cars are EVs and they are being used as primary family cars for road trips, and everything else, not just commuting), and I don't think the solutions are cheap, easy, or going to be readily available in the next decade.
That plan calls for covering 19% of the southwest with solar installations. I would say that is quite a bit larger than 22% of california.
And it only provides 69% of the electricity (oddly, only 35% of the energy... so we still need 65% fossil fuels?) I thought the idea of all this was to get off of fossil fuels by getting all cars, homes, businesses using clean electricity for heating, cooling and transportation.
So, we'd need to cover 25% of the southwest (not just california) for 100% of current electricity. Unfortunately they don't take into account the instant 25-40% increase in electricity usage that will occur when we all plug in our EVs for the first time.
Also, it is talking about PV. solar2 and solar tres are not PV installs, they are reflected light focused on a central tower. So, my question still stands, does the solar tech in the article not scale?
sorry to reply again, but thought of something else.
Even if lithium batteries increase their capacity by 10 times, there is still the problem of the life of the battery. Toyota recommends that you change the battery in the prius every 2 years, at a cost of $4500. Well, that $4500 dwarfs my gasoline costs for 2 years even without the power cost. This hidden cost is rarely if ever discussed when talking about EV or hybrid vehicles, but it is there. It is the reason I didn't buy a prius last year when I was shopping, the dealership was very up front and open about this cost, and stated that after 2 years the battery is almost useless and that the car would run almost 100% on gas unless you replace it. This cost makes any EV or hybrid vehicle completely uneconomical.
New lion batteries have not addressed the aging issue. My 12 month old laptop has seen about a 70% decrease in battery life. When cars are experiencing this phenomena people will not put up with it, and the batteries packs in cars will cost 8-10k once they are big enough to provide 400 mile range.
While I would love an EV and I think my driving patterns would be perfect for it (about 30 miles a day, no problem plugging the car in and letting it charge for the night), I would still have to keep a gas car for road trips (I take 1-2 1000 mile road trips for work per quarter).
Further, you don't address the problem the other poster brought up. 240V 1325Amp circuits are not going to happen, they are especially NOT going to be installed in anyone's home (instant death to anyone who accidentally touches that outlet). That is what you would need for a 10 minute charge of the tesla which only has a range of 221 miles. You HAVE TO HAVE at least 1 240V 1325Amp circuit and it simply isn't going to happen. If you double that range to 440 to make it competitive with gasoline powered cars, sure the battery tech might be here soon to support that, but you still need 2 times the energy. So now you need a 240V 2650Amp circuit to get a 10 minute charge?
As for "gas station" charging stations, figure 16 "pumps", 318kw * 16 pumps = 5MW of peak power. No gas station on the planet has a line running into it that can supply 5MW of power. Upgrading the electrical grid to support this kind of distribution will cost many billions of dollars. It is like an upgrade from copper to fiber in telecom. You need new transformers everywhere, new lines, new termination equipment at the customer prem, new equipment at the customer prem to be able to supply 16 circuits capable of carrying 318kw without melting. Sure it might be cheaper than upgrading the oil distribution infrastructure... but not by a huge percentage. It isn't going to be "cheap". Again multiply everything by 2 once the batteries are here which will actually support comparable range. So now you need a 10MW line running into every gas station to support 16 simultaneous 10 minute charges.
Also, I'll nitpick here a bit, it takes me about 3 minutes to gas up my 10 gallon car (the car gets 35 miles/gallon so 350 mile range). Taking 3 times as long to fill up is probably ok for a little commuter car, what about a family car like a minivan or small suv? They will obviously need a lot more power. My parent's SUV takes ~12 minutes to fill with gas, 3 times as long is 36 minutes. I don't know anyone who will stop a family vacation with kids in the car for 40 minutes to fill up with gas.
Also, what do you do about "running out of gas" in an EV? This will happen, it happens with gas, it will happen with electrical. With gas you go get a can and fill it up. Do you go buy a spare 200lb battery pack? Do you cart around a gas generator in the trunk? They will have to invent some sort of "quick charge" device that weighs less than 10lbs, can be purchased for $3 and provides enough juice for 20-30 miles (IE a gallon gas can replacement).
Umm... Not to burst your bubble, but that calculator says that it would take 34 years to pay back a 3KW system in Las Vegas, NV..
Unfortunately the cells only last for 25 years. That is not breaking even, because 9 years before you realize a return on your investment, you have to make the investment again. If you can't break even in Las Vegas, you can't break even anywhere in the US... well maybe death valley?
writing off a factor of 3 or 4 is still bizarre. I agree the original projection of 30 times was exaggerated. However, my electrical bill is ~$100/mo, you freely admit that if we moved to solar my bill would go to $300-400/mo... That is 100% unacceptable.
Just read the wikipedia on solar2, it said they are building solar tres in spain based on the design of solar2.
It says solar 2 is a 10MW plant, and that for solar tres they are DOUBLING the size (in area of mirrors) yet it will only be a 15MW plant... So... does this tech completely not scale? A doubling in size only creates a 50% increase in production? Double again and you only get a 22.5MW plant and now you're using 4 times the area of a 10MW plant?
I'm just saying, I doubt your straight linear math works, I bet it is a whole lot more than 22% of Cali to power the US because apparently this doesn't scale.
From a later post, I guess Windows Home Server *automatically* backs up clients that are connected to it, so you have your pretty little PC you open up your pictures and crop them... and whatever, Home Server at some point will back that up (corrupting the files in the process).
Google Reader begins sharing public data in a new way.
These were not "private" feeds, they were publicly available URLs (although obfuscated).
I'm not necessarily siding with Google on this one. I do think they should have thought this change of functionality out a little more, but the fact remains this data was already public. Comparing it to the Beacon scandal is not accurate at all.
yeah I'm retarded, $5k/kW sorry for the really poor math. The quotes I've received are ~40k including installation
and it would still be probably $10k-15k after installation for these new panels too.. installation, mounting, labor is ~half of any solar install from what I've seen.