English Language & Usage: Multi-Layered Discourse Room

« first day (3642 days earlier) ← previous day next day → last day (1577 days later) »

tchrist

22:03

Holland had over 11,000 new cases yesterday.

@Robusto We had 2400 yesterday, and I think 2900 today.

Cerberus

@Robusto But if the snow cover remains intact?

@tchrist But some of those were delayed registrations from earlier.

We may be plateauing.

tchrist

@Cerberus Why would the snow remain intact? That really isn't very much snow, what with high winds, high solar gain, and actual sublimation at this altitude. It needn’t wait to melt you know just to disappear.

@Cerberus Still reminds me of April in New York. Whatever its steepness or gentleness, the new-cases slope maps almost always to the new-deaths slope a month later.

Faheem Mitha

@M.A.R. No I haven't. I don't think I'd even heard of it.

tchrist

I believe England is going into some sort of lockdown for a month now. Unclear how strict.

Cerberus

@tchrist I don't know, if there is enough?

@tchrist Well, that isn't really the case here.

We have lots of indicators.

Positive tests are only one.

Faheem Mitha

22:13

@CowperKettle Are you on your way to a totalitarian dictatorship, or are you already there? And for what it's worth, I'm sorry.

Cerberus

Sewage is the earliest indicator, then tests (which is somewhat unreliable), then hospitalisations, then IC, then deaths.

Those two sudden drops were caused by computer problems.

As is the suddenly high peak.

tchrist

@Cerberus Yes, sewage is a leading indicator.

Faheem Mitha

@Robusto I didn't realise conditions in Thailand were so bad. It's normally mentioned as one of the few places in the region which isn't a complete basket case. In part because the British, never, er, visited.

Cerberus

At least the reliable indicators are rising much slower than in March.

@FaheemMitha I don't think that is fair.

There are plenty of former colonies which are doing comparatively well.

Like Malaysia, Singapore, Hongkong, Taiwan, Korea, Vietnam...

tchrist

How is your positivity rate?

Cerberus

22:16

Probably high.

tchrist

From nytimes.com/interactive/2020/us/coronavirus-testing.html and updated for today.

South Dakota has a 43% positivity rate. Wisconsin 14%. Colorado is 6% overall but 7% or 8% in some of the harder-hit counties.

Averaging across such huge areas hides how much worse it really is some of these places. Hospitalization rates and availability, ICU ditto, etc.

Cerberus

Yes.

One of the most important figures, in my opinion, is the estimated number of infectious people.

Faheem Mitha

@Cerberus It's entirely fair. and they're mostly fairly small places. And whether they are doing well, I don't know. Both Singapore and Vietnam are basically police states. North Korea is cut off from the world.

tchrist

22:32

@Cerberus Important to or for what purpose?

1 hour later…

Cerberus

23:35

@FaheemMitha All of those are doing well enough, compared with Thailand.

tchrist

Seems like a poor day to throw a party with a hundred and thirty thousand people jammed together, doesn't it?

Cerberus

@tchrist Important to characterise how 'bad' the current wave is.

tchrist

Or maybe not! Hugs, sequins and rainbows as Taiwan enjoys victory over coronavirus

Cerberus

They may or may not regret it.

I'm sure the government has made a fair assessment.

tchrist

It was their two hundred and second day without a locally transmitted case of covid.

Faheem Mitha

23:38

@tchrist Can you comment as to whether there is any reason to use the Python 3 re module vs PCRE in Python 3? Though that may be a moot point, since Debian doesn't seem to have a PCRE for Python. Which seems surprising, so maybe I need to look harder.

tchrist

@FaheemMitha Maybe?

just a sec

@FaheemMitha This is the one I used to recommend for Python, but it has been so long now I don't know the current landscape’s perimeters.

But it does appear to be under active support.

Faheem Mitha

@tchrist This includes PCRE?

tchrist

No.

But for that there's always Perl. :)

What it does include is proper Unicode support.

You don't have to be stuck in your grandparents’ all-ASCII world. :)

Faheem Mitha

No, doesn't look like it.

I was hoping, expecting, really, that Python would have a standard/maintained PCRE library.

tchrist

It has many PCRE features.

It doesn't. They don't like us.

Faheem Mitha

23:43

But perhaps there's some NIH thing going on.

tchrist

Well.

Faheem Mitha

@tchrist Bummer.

tchrist

It's complicated.

There were at least historically a lot of vocal Python converts who've been screwed over by being forced to maintain poorly written Perl code. This of course makes as much sense as cursing the hammer and screwdriver for a poorly built house or the paints and brushes for a poorly drawn portrait.

Bad code is written by bad coders, not by a bad programming language.

Faheem Mitha

I was hoping to standardize on PCRE, since it seems to be something of a standard. Or at least, as much as a standard as prevails in this area. Which doesn't seem to have actual standards.

@tchrist I don't suppose that's packaged for Debian.

tchrist

@FaheemMitha I have no idea. I just download tarballs and type sudo make install. :)

Faheem Mitha

23:48

Looks like it was there, and got thrown out...

@tchrist I like binary packages.

tchrist

When you say "standardize on" PCRE, what do you mean? Do you mean across all possible programming languages and compilers and operating systems? That seems a tall order to fill.

POSIX has two flavors of "standard" regexes, but they define too small a feature set for anyone to want to use just those alone.

I would also recommend Russ Cox's re2 library that he originally wrote for Go.

Faheem Mitha

@tchrist Yes, that's what I meant. Though my possible use cases would be quite meagre, since I only know a few languages.

No, it's there after all. I even have it installed, for Python 2. For some reason apt-cache search python regex doesn't pick up python-regex. :-) I guess I need to install the Python 3 version now.

I wonder if one can get regexes for RDBMS.

And if so, whether they are useful.

tchrist

I would normally recommend that one measure a regex implementation against conformance with Unicode Technical Standard #18: Regular Expressions, but perhaps you still use only TELEGRAM COMPATIBLE CHARACTERS.

@FaheemMitha Sure, mysql support REGEXP LIKE.

Faheem Mitha

@tchrist Yes, I just use English, and ASCII.

@tchrist So Barnett's regex library is preferable to re overall? And is it a strict superset in terms of features?

tchrist

@FaheemMitha Not so much. RDBMS are simply shit for character work. They don't optimize for text searches.

Faheem Mitha

23:55

@tchrist Ok.

tchrist

@FaheemMitha It's been almost a full decade since I last dove into that python world, so I cannot say.

Faheem Mitha

@tchrist Ok. Thank you for your guidance and suggestions.

@tchrist So the regex library is "close" to PCRE overall?

tchrist

Basically, searching text requires a special kind of backing-store indexes to be built, and very few RDBMs provide that sort of thing. I believe Oracle may.

@FaheemMitha That was my impression long ago, yes.

Faheem Mitha

@tchrist I see. Anyway, it's not important. I have no concrete plans to do such a thing.

tchrist

I do on occasion use REGEXP LIKE in mysql, because LIKE sucks COBOL’s burro detective.

Robusto

23:59

OK, new weirdest swype error on my phone: I swyped "somewhere else" and it came out "Stonehenge eels." And I didn't even know Stonehenge had eels.

tchrist

What is swype?

Faheem Mitha

It looks like I'll have to rebuild the most recent version of python3-regex currently available in unstable for buster, because it won't install on it. Assuming that's possible, of course.

tchrist

Drawing letters?