That miss was what pushed me to google podcast in the first place, along with their absolute refusal to fix broken feeds for existing podcasts in their index.
Really frustrating to see the amount of effort that went in to gaming those benchmarks when catastrophic performance issues in core Microsoft-authored libraries go unfixed for years. https://github.com/dotnet/SqlClient/issues/593
It's something that is completely unintuitive because MS wants you to use as much async as possible, provides an async API but is not fixing the damn issue for years. Most senior devs will not know about this, use the methods and probably won't even notice it in most production scenarios. At least not until it's hits them like a ton of bricks.
Another favorite .NET tidbit of mine is the existence of Microsoft.VisualStudio.Threading. If you use a lot of async code with UI you will eventually run into deadlocks. MS noticed that, too, and wrote this library for Visual Studio (hence the name) to work around that and then released it to the public.
Also fun (but understandable why they don't fix it) BUT with VS2019 they included a new "code fix"/analyzer that very annoyingly tells you to convert the latter into the former for "code readability", making you code slower.
Don't get me wrong, I like .NET a lot but it has quite a few footguns.
> It does pretty much what you say: it repeatedly copies data (2 bytes at a time though), without materializing a string. So if the string you try to read is 10 packets long, you end up copying 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + 10 = 55 packets. If you need 100 packets, you end up copying them 5050 times. So to read N bytes you copy O(N^2) bytes: it's impossible to scale this way. Whether bytes are copied 2 at a time or using a more optimal algorithm doesn't really matter.
Good lord. It took them about a year to find this obvious O(n^2) hot loop and they’re unable the fix it.
To me it seems like the people writing that parser are just doing it piecemeal, adding little spot fixes instead of having a consistent and robust stream parser that won’t crash or slow down to molasses if you look at it wrong.
In several places they mention that the parser was never tested with partial buffer fills because replayed packet captures from disk don’t do that. Real network sockets do.
Suddenly I can see how the security bug happened where data was getting received by the wrong SqlConnection under high load.
I bet they have a bunch of concurrency bugs in that spaghetti code…
> For similar-sized vehicles in the U.S. today,
per-mile lifecycle (including vehicle and battery
production) greenhouse gas emissions for
battery electric vehicles run on the present
U.S.-average grid electricity are approximately
55% of the emissions from conventional
internal combustion engine vehicles.
Electric vehicles have almost half the lifetime emissions. That's huge, no matter how you spin it.
Sure but with about 167% as many emissions from production, I'm just saying there's going to be an immediate increase in emissions.
Also one thing missing from this analysis is maintenance. The three factors taken into account are initial production (much higher for EVs), cost of fuel production (also much higher for EVs), and emissions from fuel combustion (0 for EVs). However batteries don't last forever and it still remains to be seen that EVs can reach the maintenance costs of combustion engine vehicles (with Tesla being a notable argument against EVs potential)
Fuel production is much cheaper for EVs, what are you talking about?
The batteries last longer than the 150,000km these studies assume. My 2013 Model S is at 209,000 kilometers and going strong with little (10% or less) range reduction.
I looked at the study a bit, and they use Model S-sized battery size and efficiency assumptions (about 3mi/kWh) to compare to a Camry. More appropriate would be Model 3 assumptions (4mi/kWh). The emissions factors are from a white paper 5 years ago, itself using older data. They assume 525grams of CO2 per kWh for the reference case going down to 345 in 2050, but the US already has emissions of about 375gramsCO2/kWh and falling. Just terrible assumptions. A bunch of stuff like that in the study.
And it compounds! A factor of 1.33 bigger battery (3mi/kWh instead of 4mi/kWh) whose manufacturing emissions are 1.41 times as high (525grams of CO2 per kWh vs 372, if we optimistically assume electricity is the main energy input but pessimistically assume the energy needed to make a kWh of capacity remains the same) means a factor of 1.9 exaggeration in manufacturing emissions. Plus the operating emissions per mile are also exaggerated by a factor of 1.9…
Finally: Batteries in modern EVs last the life of the vehicle. 500,000km or so. Potentially longer with LFP cells.
> Fuel production is much cheaper for EVs, what are you talking about?
No it's very much not. Please see chapter 4 of the linked MIT report:
Fuel production emissions are also
typically higher for BEVs (and FCEVs) because, on
average, generating and delivering a megajoule of
electricity or hydrogen to a vehicle battery or fuel
cell consumes much more energy than producing
and delivering a megajoule of gasoline to the fuel
tank of an ICE
But you said “cheaper,” additionally this ignores the input fossil energy of the gasoline, and it’s per joule of thermal energy, not useful mechanical energy. It’s a weird metric that isn’t very enlightening. It’s mixing low-entropy electrical energy with high-entropy thermal “primary energy.” It’s also not what you actually said. The cost of energy (as well as emissions) per mile traveled is far, FAR less in electric vehicles.
Again, you're misunderstanding the point. I really recommend you just check out the first few pages of chapter 4 of the full MIT report. Yes the emissions per mile travelled is less in EVs. Nobody is arguing against that. There's 3 categories here:
1. Emissions from initial production
2. Emissions from fuel production
3. Emissions from fuel consumption
For EVs, emissions from 1 and 2 are higher. But this is more than offset by having 0 emissions from the third category. So yes, the emissions per mile traveled is less in EVs. However the cost of fuel production per mile traveled is still significantly higher (around 192% higher)
PS you're the only one who used the word "cheaper". I only used it when directly quoting you. I presumed we were still measuring by emissions and not by dollar cost, but I see now that you were talking about something else
- Special Counsel Mueller declined to exonerate President Trump and instead detailed multiple episodes in which he engaged in obstructive conduct
- The investigation “identified numerous links between the Russian government and the Trump Campaign” and established that the Trump Campaign “showed interest in WikiLeaks's releases of documents and welcomed their potential to damage candidate Clinton”
- Russia engaged in extensive attacks on the U.S. election system in 2016
- The Special Counsel investigation uncovered extensive criminal activity
> instead detailed multiple episodes in which he engaged in obstructive conduct
Yes, obstructive conduct. Not collusion with the Russians.
The very fact that Mueller detailed episodes of obstructive conduct and not episodes of collusion reveals how little evidence there is to support allegations of collusion.
And as for the obstruction, remember that an FBI lawyer pled guilty to doctoring an email that was submitted as part of a FISA application used to surveil Trump campaign adviser Carter Page.[1] And Igor Danchenko, the primary researcher of the infamous Steele Dossier, is awaiting trial for lying to the FBI about the dossier's sources.
So is it really a crime to "obstruct" an investigation that is based on lies? Such an investigation should be shut down.
That miss was what pushed me to google podcast in the first place, along with their absolute refusal to fix broken feeds for existing podcasts in their index.