Paul Evans

https://fosstodon.org/@leonerd

Perl, Electronics, General geekery.

Pronouns: he/him
Twitter: cpan_pevans
Github: https://github.com/leonerd

Public Key

npub1hlpkkqf7sthaqvhufu9v2k8p0tpy5vcv8df6zaw4jzm36s3x3g7srrgkyk

NIP-05 Address

leonerd_at_fosstodon.org@momostr.pink

Profile Code

nprofile1qqstlsmtqylg9m7sxt7y7zk9trsh4sj2xvxrk5apwh2epdcaggng50gw7wvl5

Author Public Key

npub1hlpkkqf7sthaqvhufu9v2k8p0tpy5vcv8df6zaw4jzm36s3x3g7srrgkyk

Show more details

Last Notes

2024-09-09 18:23:43

- reply

Further update on this one: The DisplayPort also seems entirely unfunctional. It doesn't appear at all under xrandr, though the HDMI alongside it seems to be fine at least.

2024-08-21 22:07:43

#TodayIAchieved A new subcommand to my "csvtool" program, for applying smoothing filters like windowing or recursive average, across time-based data. Applying it to some sensor data makes it much cleaner, easier to see what's going on.
https://metacpan.org/release/PEVANS/App-csvtool-0.03/view/lib/App/csvtool/Smudge.pm
https://cdn.fosstodon.org/media_attachments/files/113/002/276/160/231/812/original/2f1654d39a95fdd5.png
https://cdn.fosstodon.org/media_attachments/files/113/002/277/336/235/164/original/d679a16b06980aca.png
https://cdn.fosstodon.org/media_attachments/files/113/002/277/332/103/902/original/c276c9b496f627c5.png

2024-08-20 22:56:24

- reply

Better written as `STDOUT->autoflush` if you want the #perl code to actually be meaningfully readable ;)

2024-08-19 19:25:49

- reply

Ah yes likely the wording in the message needs improving then

2024-08-19 09:14:27

- reply

It's only experimental now if you add a `finally` block

2024-08-18 22:03:37

- reply

I've often wondered why no laptop ever seems to support being used as a simple HDMI monitor by directing video sent in to the HDMI port up to its internal display panel, bypassing the local GPU entirely

2024-08-18 09:54:54

- reply

And yet they still wrapped it in four $(dirname ...) command substitutions

2024-08-16 23:30:51

- reply

Ahyes, "inf" and "nan", the unfortunate almost-a-number values

2024-08-16 11:58:58

- reply

A curious thought, but various poking at it doesn't seem to make that work either.

2024-08-16 11:43:39

My string of disappointment with USB=C docks continues. Today it's the #Belkin 11-in-1. It reliably acts as a USB A hub and ethernet, seems to do passthrough power just fine, and my minimal testing of the HDMI port has no complaints, but the audio port is entirely missing. Not just "unreliably works", but I don't see anything listed on USB that could possibly be it, at all. It's as if the thing does not have an audio port at all.
https://cdn.fosstodon.org/media_attachments/files/112/971/519/165/922/571/original/5181217c25ffc3ad.webp

2024-08-15 21:26:28

- reply

Actually fun fact: Go's near-equivalent didn't do that either until fairly recently

2024-08-15 21:26:01

- reply

Python doesn't have variable capture from outer scopes, or closures; thus actual anonymous functions would be far less useful than you'd imagine.
Compare:
```perl
my @closures;
foreach my $x ( 1 .. 10 ) {
push @closures, sub { return $x };
}
```
each of the ten closures pushed to the array captures the value $x had been when it was created. Invoking each would return its own captured value. Python's one doesn't do that.

2024-08-11 19:10:03

- reply

Isn't that an indication of imminent nuclear war if it did happen?

2024-08-10 16:12:26

- reply

The other "golden gun"

2024-08-08 14:30:46

- reply

Ahah; turns out to be a known bug, already fixed in 0.143 (from unstable)
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1076561

2024-08-08 13:46:28

So, #debian, in its infinite wisdom, has decided that an initramfs 247MiB in size is what it wants. Well, OK..? But now I can't fit two of them plus kernels etc on the 500MiB /boot that debian also created. Much disk shuffling ensues or my laptop is non-bootable.
I can't help thinking this a bug. That's an enormous size for initramfs, surely? How big is anyone else's?

2024-08-07 07:50:16

- reply

Yup. "Your old code still works" is a hugely important feature for a language, and one we tale very seriously.

2024-08-07 00:58:43

- reply

We're discussing it right now:
https://www.nntp.perl.org/group/perl.perl5.porters/2024/08/msg268628.html

2024-08-07 00:42:15

#TodayIAchieved Parsing of a table notation in Pod documents in App::sdview. This uses notation similar in style to Markdown. It's an idea we're experimenting with.
https://cdn.fosstodon.org/media_attachments/files/112/917/951/117/821/255/original/3f22d6fbea22a592.png

2024-08-06 18:40:04

- reply

That was indeed a change that Raku made.
It would be too difficult to change that in Perl though because of all the existing code. Since Raku (Perl 6 at the time) was a new language with no existing code, that was a thing they could much more easily do.

2024-08-03 13:02:56

- reply

And there I was expecting to see some birds

2024-08-03 10:46:15

- reply

I still think David Copperfield (the magician) should write a book titled "Charles Dickens"

2024-08-01 20:09:54

- reply

So far I've bought/tried/returned one that worked fine as a USB-C to A hub and ethernet port, but whose onboard HDMI and audio outputs did not appear visible at all.
I'm also currently trying out a Dell one that seems fine for USB-A, ethernet, and HDMI. It has audio that doesn't even appear as a USB device until you unplug/replug the audio connector in, every time. Most annoying. Also half the times I plug it in it doesn't do power passthrough.
My experience thus far has not been good.

2024-08-01 20:06:17

I have myself a new USB-C laptop. Which means I'm in the market for a USB-C dock. One that's actually any good.
The ones I've tried so far have all been far from impressive.
Before I try buying yet another disappointment, does anyone have a suggestion on a good one? By which I mean, one that ... actually.. works. In that, all its included features actually perform correctly, reliably.

2024-07-31 19:40:50

- reply

We have that in #perl, with my Devel::MAT. In fact I even named it based on the `jmat` tooling I think you're referring to:
https://metacpan.org/pod/Devel::MAT

2024-07-31 12:50:44

- reply

Though it appears to be using Text::Balanced at least, so it's perhaps not quite as bad as it might otherwise be

2024-07-30 20:09:47

- reply

There's a mistake you don't make twice ;)

2024-07-30 19:05:38

As a programmer is it important to know the difference between an API and an IPA. Misusing either can mess up your evening, but the solution to each is very different.

2024-07-09 08:52:53

- reply

:Checked(CodeRef)
with Signature::Attribute::Checked + Data::Checks

2024-07-08 21:32:59

Someone, tell me: How am I supposed to set up a server, so that mail to its `root@`... account actually gets sent to somewhere I can see? Right now it's just all sitting in that machine's /var/spool/mail. Kinda useless. I want it where I can see it.
What do I have to install and set up and why isn't that done as standard on a debian box?

2024-07-08 21:31:45

- reply

That is indeed the entire point of it yes ;)

2024-07-08 13:18:54

- reply

And now fixed:
https://metacpan.org/release/PEVANS/Feature-Compat-Class-0.07
Thanks for reminding me.

2024-07-08 11:19:35

- reply

Oops, I haven't got around to adding that one yet; should do that today. I'll let you know...

2024-07-06 10:30:33

- reply

They're very layerable - I do it all the time ;) See my other reply:
https://fosstodon.org/@leonerd/112739075226480192

2024-07-06 10:29:27

- reply

Unsure about that (also I presume you meant Object::Pad).
Object::Pad does have an annoying bug where some syntax error messages get eaten by the perl core and lost, so all you know is that an error happened, but not where.
But in the absence of errors, it's very robust - I often use it in a big combination with Future::AsyncAwait, Syntax::Keyword::Match, and various other keyword/infix modules.
For example, here's an `async method` with match/case in it:
https://metacpan.org/dist/Device-Serial-SLuRM/source/lib/Device/Serial/SLuRM.pm#L387

2024-07-06 10:25:22

- reply

Don't forget of course there's `Feature::Compat::Class`, which uses the latest core feature on latest perl, or uses Object::Pad to polyfill it in on earlier perls. It is restricted to only permitting the same things that core would support, so you know you're writing code that'll work there even on earlier perls - right back to 5.18 if necessary.
https://metacpan.org/pod/Feature::Compat::Class

2024-07-05 23:03:00

- reply

You can see it in action here:
https://fosstodon.org/@leonerd/111750856436353593

2024-07-05 23:00:21

- reply

If you find those interesting, you might enjoy my `sdview`, as it can render POD and manpages (in addition to a few other formats) as well as do some cross-conversion.
https://metacpan.org/dist/App-sdview/view/bin/sdview

2024-07-05 19:17:00

- reply

Yup - how do you suppose all the perl*(?) pages get created anyway? ;)

2024-07-05 11:20:48

- reply

Maybe? But that wouldn't explain the problem that I have a cable that used to work but now does not. Nothing else has changed besides some wear. It's the same cable between the same endpoints. Worked last week. Does not work today.

2024-07-05 11:10:22

Does anyone know of a USB A->C cable tester that actually stress-tests the power-carrying conductors for resistance? I see some that just check yes/no to connectivity. But that's not sufficient for getting good charge.

2024-07-04 11:27:57

- reply

Pepperidge Farm remembers.

2024-07-03 09:13:05

- reply

Seems good, yes

2024-07-01 22:37:14

- reply

And if anything else is missing and needs writing, there's always
https://metacpan.org/pod/Net::Prometheus

2024-07-01 12:18:04

- reply

I have about 4

2024-07-01 10:38:36

- reply

It's not. It's weird. Indirect notation was (a bad idea) invented after that, to sortof copy the same notation.

2024-07-01 10:23:16

- reply

A whole bunch of them are parser-special tricks - `print` in particular is weird because of the indirect filehandle like that.
Much better in that case to write it in method form, though:
```
$file->print( $foo, $bar, ... );
```
makes it much more obvious to readers what is going on.

2024-07-01 09:42:02

- reply

There wouldn't be much use for it. A key point about Perl is that you can always run a later version of the perl interpreter and it still copes with older code. That's what the back-compatibility means. If you have a program that needs 5.40, then just run it with 5.40. Anything built for older perls still works.

2024-06-28 10:09:22

- reply

Planned, specified, mostly implemented even. Hopefully we can get this in for next year's release
https://github.com/Perl/PPCs/blob/main/ppcs/ppc0021-optional-chaining-operator.md

2024-06-28 09:12:15

#TodayIAchieved A new version of my recently-new #CPAN #perl module `Data::Checks`, which adds some parametric constraint checks.
New here is `Isa`, which accepts an object reference only of some given class, and `Maybe`, which turns another constraint check into one that also accepts `undef`.
#perl
https://metacpan.org/pod/Data::Checks
https://cdn.fosstodon.org/media_attachments/files/112/693/465/212/279/511/original/298dfe6804b27139.png

2024-06-26 11:16:24

- reply

Inkscape lets you type numerical expressions (such as 123 + 45/6) into any textbox that takes numbers, and I always get annoyed that KiCAD does not

2024-06-24 09:20:41

#TIL that `ssh` doesn't implement HappyEyeballs [1]. If it thinks it has IPv6 connectivity but the upstream router doesn't work, it'll just sit there never making progress, never failing back to 'v4 instead. You have to manually `ssh -4` or `ip -6 ro del default` to get around that bug.
[1]: https://datatracker.ietf.org/doc/html/rfc6555

2024-06-24 07:48:15

- reply

Still looks surprisingly cute. Would pet. 8/10.

2024-06-22 07:48:04

- reply

Yes, but start with something new and modern. `use v5.36` and so on... A lot of the new bits are much nicer than "the old stuff"

2024-06-21 20:32:34

So, alcohol cannot be consumed anywhere at all.
Except perhaps on the zero-width border that defines the exact perimeter of these premises.
https://cdn.fosstodon.org/media_attachments/files/112/656/504/460/597/461/original/3100f94ee7923ea8.jpg

2024-06-20 20:34:40

- reply

You're basically looking for what I was trying to do with Circle then

2024-06-17 12:35:24

- reply

I've tried thinking about it but in practice it'a hard to find syntax any nicer than `sub`. It's actually quite nice already once you have signatures
```
do_thing_with sub($x, $y) { ... }
do_thing_with -> $x, $y { ... }
```
It's only very slightly longer to type but is already valid syntax, and pretty clear in meaning. I don't think there's much to be improved upon here.

2024-06-15 22:51:24

- reply

Little FYI: There's no need to `use strict` or `use warnings` when you already have a `use v5.36` in scope. They're already implied.

2024-06-15 14:05:29

- reply

https://github.com/tree-sitter-perl/tree-sitter-perl?tab=readme-ov-file#neovim

2024-06-15 13:39:13

There's a kind of cyclic-dependency on the new `meta` module for #Perl, in that until it's in core I suspect a lot of folks won't use it, but until more folks use it so we know it's any good, we won't put it in core.
What we need is some more volunteers to actually try using it for something real. I have a few uses of it already, but it'd be great to see more modules from more authors than just me on this list:
https://metacpan.org/dist/meta/requires

2024-06-15 08:47:11

- reply

ferrets! Seems like quite a *dark shades* business you have there

2024-06-15 08:43:28

- reply

A nice example, though it also shows off how much vim's syntax highlight isn't keeping up with the new updates. You may get better highlights in your examples by using neovim and its tree-sitter + tree-sitter-perl

2024-06-11 16:20:06

- reply

Oh indeed, I don't imagine I'd really ever use it either, I just added it because it was an easy fix that makes the set complete. Without it there was an awkward gap.

2024-06-10 22:01:21

- reply

Every release has a "perldelta" document that lists the changes since last release. You could read those in order since your last one.
Here's the latest, for example:
https://perldoc.perl.org/perl5400delta

2024-06-02 17:31:03

- reply

Perhaps, if there isn't, I should consider writing this as a bug to XFCE4 itself? It's a pretty simple fix - xfce4-session can detect the presnece of a `.profile` in the user's home and reëxec itself via that little shell fragment if it finds it there.

2024-06-02 17:29:45

I run #XFCE on my Linux box. I have a custom `.profile` that needs to be run to set $PATH and a bunch of other environment things. It seems by default xfce4-session doesn't do this.
I can workaround that ;problem by this file:
```
$ cat $HOME/.xsession
#!/bin/sh
source $HOME/.profile
exec xfce4-session
```
and remembering to set "default X session" rather than "XFCE4" as the login environment.
Seems incredibly hacky though. Is there anything nicer?

2024-06-01 19:24:03

- reply

"a nearby Tesla coil is causing phantom key presses on out giant recreation ZX spectrum" is very peak-EMF

2024-05-30 13:56:00

- reply

In installments. Over a series of days.

2024-05-23 08:39:33

- reply

Ahyes, very fair

2024-05-23 07:02:28

- reply

You forgot to start at the oddball 110

2024-05-20 15:47:57

- reply

This isn't about shells. This is about the xfce4 session overall.

2024-05-20 15:47:29

- reply

Yeah I could do that, but then if I do nothing else then X just exits back to login screen. So I additionally have to `exec xfce4-session` at the end, which feels a bit ugly, but perhaps that's OK.

2024-05-20 15:07:32

I have a `.profile` that sets up PATH and some other environment variables. It applies fine when I start a new ssh login. But it doesn't appear to take effect for my X session, so window manager etc.. can't see the extra PATH entries.
What's the solution here? How do I ensure X desktop (XFCE in my case) actually sees it?

2024-05-20 08:34:04

- reply

If you want decent integration with 3rd party stuff (google APIs, amazon, etc...) you may need to write your own client stuff as most big service providers seem to have forgotten that #Perl exists

2024-05-16 16:27:02

- reply

So what you're saying is these businesses expect to be able to take stuff for free then charge money to give it to other people? Sounds like the existing oil, chemical, minerals, etc... industries to me.

2024-05-06 21:24:44

- reply

Show me a person who says they actually like computers and I'll show you a person who clearly doesn't understand enough about computers.

2024-04-09 11:26:27

I'm old enough to remember when the "Back" buttons in web browsers used to work reliably. You'd get exactly one page transition back in history for every press of the button.
Can we bring that back, please?