> My understanding of how IRC works is that you miss everything that is
> sent while you are not connected. Can you speak more to a
> "recommended" set-up such that if you are not around, you can still
> easily find highlights when you go to look?
>
> Is the answer to set up a bouncer? Is simply that your computer always
> has a client running? Does your client mark highlights for you to easily
> view at your leisure?

Exactly my thoughts, thanks for spelling them out.

In addition to these, I'd add a note about SSL encryption and end-to-end
encryption (or lack thereof), what the expectation should be in

"Alexey Yerin" <yyp@disroot.org> writes:

> --- BEGIN CONFLICT OF INTEREST ---
> * https://sr.ht/~yerinalexey/sushi
> * https://sr.ht/~yerinalexey/gtranslate
> --- END CONFLICT OF INTEREST ---

I'd like to double down on your self plug suggesting also this:
https://sr.ht/~yerinalexey/rex/

I didn't use it yet but I find it interesting. I'll keep it in mind next time I need it :)

> If you click the Refs tab on https://git.sr.ht/~gpanders/ijq and then click on
> the name of the most recent release (v0.3.6 in this case) you should see a link
> on that page that contains a compiled binary release. Note that at this time,
> binaries are only compiled for x86_64 Linux.

Alright, got it, thank you so much.

I've briefly tested the tool and it's brilliant, great work!

Hello,

perhaps I'm not too familiar with how sr.ht stores binary releases, but I
could not find a way to download the binary release (i.e. with the `ijq`
file) by reading the README.md.

This link seems to point to the source release:
https://git.sr.ht/~gpanders/ijq/refs/vX.Y.Z/ijq-vX.Y.Z-linux-x86_64.tar.gz

Can you help me out? Am I missing something? :)

thanks

Nik writes:

> I have a maybe naive question. Out of simplicity principle, why 
> not to
> treat ML algorithms in exactly the same way as human learning. 
> So if
> something is public other people already _can_ learn from it and 
> use in
> their work. We already have "models" in our brains. What's the 
> principal
> difference here?

My hunch is that human learning cannot be assimilated to machine

Rust has a three tier suppot level for various architectures [1]. I am
unclear which `aarch64` you refer to, but if it's a Tier 1, it must work
and you should probably open an issue.

Other arch are a "best-effort" so support depends if there is a
volunteer with the proper hardware.

In any case, I'd suggest you to open a bug (or check if there is one
already opened).

cheers,

[1] https://doc.rust-lang.org/nightly/rustc/platform-support.html

Hi Drew!

interesting your latest on offline mail management, I'm also a fan of
mbsync + $(MY_MUA). A tiny nit from my config:

# force recent TLS, by default mbsync tries everything fron TLSv1 to
# TLSv1.3 - not sure how many TLSv1.3 around
SSLVersions TLSv1.2

Though I'm curious why you setup a local Postfix instead of getting
mail directly from Migadu (agreed, Postfix is a PITA to configure). If
your goal is to have an offline copy, why not using straight good ol'
POP3(s)? I use mpop[1] for this.

Mehdi Sadeghi <mehdi@mehdix.org> writes:

> I was wondering whether it is possible to use lists.sr.ht as possible
> comment section for pages.sr.ht or a page hosted elsewhere. The idea is
> leaving a mailto: link on the page which will post a reply to a mailing
> list. Subsequently a webhook could be used to notify the static host to
> fetch replies using lists.sr.ht api and rebuild and republish the page.

If you don't abhor a bit of JavaScript in your page, you can try
embedding a bit of code that uses the Sourcehut paste API [0]. You can
code a form that sends a POST (create a comment), and then do a GET to
retrieve them. One comment, one paste.sr.ht item.

Just out of curiosity, why use the UTF-8 entity instead of the ASCII character?

Drew DeVault <sir@cmpwn.com> writes:

> On Sat Feb 27, 2021 at 3:58 PM EST,  wrote:
>> As of now 'curl https://jkoshy.net' has started working, which is
>> great. 'curl https://www.jkoshy.net' however continues to report
>> the SSL error.
>
> These are treated as separate sites, and you need to publish them
> separately. I disrecommend using www. in $CURRENTYEAR, though.

curious about recommending against 'www.': why?