anarazel

81 ms·

841.
by anarazel 8y ago
> The evidence coming from these 3 studies looks quite compelling I have to say. I like the trick to select a control group by acidity in the water.
842.
New Evidence That Lead Exposure Increases Crime (2017)(brookings.edu)
329 points by anarazel 8y ago|147 comments
843.
by anarazel 8y ago
> Does improving query optimizers or other internal stages sound interesting? If you've been a DBA for 10 years then you're going to know a few things about the way databases work internally, even if you specifically only have
844.
by anarazel 8y ago
This optimization applies only for new columns with a stable default. Therefore a table scan is not required to check for NULLs, as either the default is NULL or not.
845.
by anarazel 8y ago
> For instance, Microsoft SQL Server has a built-in public change tracking API, which you can tell is a leaky abstraction over top of the journal, Postgres' logical decoding [1] provides something very similar. [1]: https:/&#x
846.
by anarazel 8y ago
> Why not just consume the actual DB journal? It's decidedly non-trivial to do so, as usually the journal doesn't contain all the information to do so. Journal writes are often a bottleneck so DB engines try to restrict their c
847.
by anarazel 8y ago
But speechnow then effectively did, and that was basically just applying the framework of citizens united.
848.
by anarazel 8y ago
There's no 16bit number limiting query length in PG, the limit is 1GB and has been for a LONG time. But it's indeed often a good idea to use a temp table for some of such cases.
849.
by anarazel 8y ago
That's only the case if you add a DEFAULT value for the new column. Up until the soon-to-be-released v11 that required updating existing rows. Without a default it's essentially just inserting a single row in a row into an interna
850.
by anarazel 8y ago
> it would lock entire table while adding a new column It does so for the entire << 1ms it takes to add the necessary metadata.
851.
by anarazel 8y ago
> Note that exercising this claim yourself will permanently disqualify you from government clearances at all levels. I don't think this is quite true on the US side anymore. I think the newer guidelines (SEAD 4, guideline C) have re
852.
by anarazel 8y ago
You could also just have worked on lowering those overheads in PG, just saying. It's easy to blame "PG devs", but most of us could get changes quicker to our company's respective customers by just fixing everything in f
853.
by anarazel 8y ago
> The old "nobody ever got fired for buying Big Blue" argument. It's true that Postgres has been around for much longer than InfluxDB. Yes. Postgres appeals to managers the same way IBM / Oracle did / do. Right-y
854.
by anarazel 8y ago
>> One criticism of New York Times that I've read online is that they won't allow you to cancel your digital subscription through their website. They force you to call them. I'm not sure if that has changed recently, bu
855.
by anarazel 8y ago
Because "emergency fund" and "illiquid for six months" aren't a great combination?
856.
by anarazel 8y ago
For a lot of queries EAV type schemas are really hard to index efficiently. E.g. searching for something like a = ? and b = ? where a and b are dynamic attributes you can't just have a multi-column index when using EAV. So you instead
857.
by anarazel 8y ago
> Today we still see some use of process-level concurrency/parallelism, such as in Postgres. FWIW, as somebody working on postgres, there's plenty reasons to regret this. The unshared memory, fd spaces make some things more com
858.
by anarazel 8y ago
> The answers bring up the topic of updatable views, which got me curious. Even for those, INSTEAD OF triggers provide a better solution these days. > Views are great for bringing together data from multiple tables, but it seems that
859.
by anarazel 8y ago
True. But it's imo sad thing that it is so (at the very least it leads to duplicated functionality and duplicates query tree copies, but it also makes it harder to implement functionality like automatic matview usage etc). I'm qui
860.
by anarazel 8y ago
Sure, but a lawyer / Torvalds / any SME might want to hedge that their opinion is strong or not. There's a pretty darn big difference between "my gut feeling is that this approach is too complicated, but I've not an
861.
by anarazel 8y ago
FWIW, I find it helpful. Stating the level of confidence, and that's what IANAL is to some degree, is helpful in judging a statement.
862.
by anarazel 8y ago
Indeed (I was the PG guy at LSF/MM). I didn't feel ill-treated at all. Unless you count attempted poisoning via adult beverages, that is.
863.
by anarazel 8y ago
These day they just estimate the size of the resultset. It's pretty easy to come up with queries where it estimates the wrong row count until you paginate to a later page. I think the theory is just that approximately nobody clicks on
864.
by anarazel 8y ago
> > I've never met a company that uses oracle features that aren't available in Postgres > I'm sure you'r in the majority here on HN, but that's merely selection bias. Also, it's true today, but was it
865.
by anarazel 8y ago
How about reacting to what I actually wrote?
866.
by anarazel 8y ago
As I said, I don't know the right answer about the 1A claim. I suspect a factually narrower ruling would have entirely been possible. But they definitely could have ruled exactly the same way without making the insane "do not gi
867.
by anarazel 8y ago
> I wish they had gone further. In the modern age the concept of data hosted or stored with a “third party” is no longer useful in its entirety. ISTM that that's much more a task for the legislative branch than SCOTUS. Now congress
868.
by anarazel 8y ago
> 1. The article says “Roberts noted that the decision also allows for warrantless cell-tower location information searches in emergencies and for national security purposes.” FWIW, that's not how I read Robert's opinion. He sa
869.
by anarazel 8y ago
> Citizens United has got to be one of the most misunderstood decisions of the Supreme Court. > If Citizens United loses that case, then there would be little standing in the way of, for example, prohibiting the NY Times from running
870.
by anarazel 8y ago
> Each connection even an idle one consumes roughly 10 MB of overhead Citation needed :P Unless you're using a large shared_buffers setting without configuring huge pages, you should see smaller overhead. Obviously depends on wha

More ›