~ignaloidas

~ignaloidas/public-inbox

Last active 9 months ago
View more

Recent activity

Typo in "Breaking down Apollo Federation's anti-FOSS corporate gaslighting" 24 days ago

From Ignas Kiela to ~sircmpwn/public-inbox

 > The community, allegedly at the heart of everything Elastic does

I assume you meant Apollo

Re: How SmarterEveryDay's 4privacy can, and cannot, meet its goals a month ago

From Ignas Kiela to ~sircmpwn/public-inbox

One of the worries I have is how the work on foundations they must have 
already done just isn't good enough. I personally look for any new 
cryptosystem work to be at least better than Keybase, and I haven't seen 
any yet. For all it's downfalls, Keybase had a really nice cryptosystem 
with a lot of room for user-friendliness and abstraction, without 
getting out of hand with complexity. I'd hope that 4privacy does learn 
from it, but I need to see it to be certain. I did ask them to review 
their whitepaper, so we'll see how it goes.

Ignas Kiela

Re: [PATCH sr.ht-apkbuilds v2] *.sr.ht: Enable prometheus multiprocessing mode 2 months ago

From Ignas Kiela to ~sircmpwn/sr.ht-packages


On 2021-09-17 11:03, Drew DeVault wrote:
> On Tue Aug 31, 2021 at 5:19 PM CEST, Ignas Kiela wrote:
>> Using rm -rf for removing old files instead of checkpath -D because it
>> straigt
>> up doesn't work(https://github.com/OpenRC/openrc/issues/335)
>> ---
>>   
>> start_pre() {
>> checkpath -f "$LOGS" -m 644 -o "$BUILDS_USER:$BUILDS_USER"
>> + checkpath -d "$prometheus_multiproc_dir" -m 775 -o
>> "$BUILDS_USER:$BUILDS_USER"
>> + rm -rf "$prometheus_multiproc_dir"/*

Re: [PATCH sr.ht-apkbuilds v2] *.sr.ht: Enable prometheus multiprocessing mode 2 months ago

From Ignas Kiela to ~sircmpwn/sr.ht-packages

On 2021-09-02 14:40, Drew DeVault wrote:
> Can you summarize the motivations for this change?
> 
Right now reported metrics are inaccurate as they only represent metrics 
for a single worker, and it's not consistent on which worker answers the 
requests. With this change, the metrics will be aggregated between all 
workers. One thing to look out for is that this mode doesn't work with 
Info or Enum metrics, and some limitations on Gauges are placed, but my 
code searches show none of them are currently used in sr.ht services.

[PATCH sr.ht-apkbuilds v2] *.sr.ht: Enable prometheus multiprocessing mode 3 months ago

From Ignas Kiela to ~sircmpwn/sr.ht-packages

Using rm -rf for removing old files instead of checkpath -D because it straigt
up doesn't work(https://github.com/OpenRC/openrc/issues/335)
---
Took me a good while to get back on this. Tested in a VM with meta.sr.ht,
everything seems to work, changes for other services are the same so shouldn't 
have any problems with that. One thing I'm not sure about is how the upgrade is
going to go, as this changes files which are potentially editied by
administrators.

Should need a careful review, lots of repetitive changes to many files, which is
prone to errors. Everything seems OK to me, but one pair of eyes isn't enough
for this.

 sr.ht/builds.sr.ht/APKBUILD                     | 14 +++++++++-----
[message trimmed]

Re: [meta.sr.ht] JSONDecodeError 6 months ago

From Ignas Kiela to ~sircmpwn/sr.ht-discuss

On 2021-05-25 13:56, Éloi Rivard wrote:
> Hi. I installed the meta.sr.ht-0.53.30-r0 package on Alpine 3.13 but
> the service does not start.
 >
> Did I miss something?

Any logs from the api service? I think they should be in
/var/log/meta.sr.ht-api.log

Re: meta.sr.ht: Signature did not match digest 6 months ago

From Ignas Kiela to ~sircmpwn/sr.ht-discuss

On 2021-05-21 15:36, Sol Fisher Romanoff wrote:
> https://mirror.sr.ht has the newer version for x86_64, but it does not
> appear to exist on sr.ht-apkbuilds. How do I bump it to the newest?
> 

Bump the version in the APKBUILD and regenerate the checksums with 
`abuild checksum`. The builds on the mirror have their versions bumped 
automatically, but this isn't committed to the repository.

Re: More Intuitive Labels In Tree 7 months ago

From Ignas Kiela to ~sircmpwn/sr.ht-discuss

On 2021-04-21 04:57, Nathan Guerin wrote:
> It's probably worth asking whether browsing a source code tree is
> something someone who isn't very computer literate will do often. My
> guess is that browsing a source code tree assumes some sort of facility
> with using a computer. If they're browsing the source tree to learn
> more, then teaching them that '..' means "cd up the directory tree" is a
> good opportunity to learn something new imo.
> 

It doesn't have to be source code, you can use git to host a lot more 
things than that, and I think giving a hint towards what it does would 
be useful.

Re: More Intuitive Labels In Tree 7 months ago

From Ignas Kiela to ~sircmpwn/sr.ht-discuss

On 2021-04-21 04:00, Drew DeVault wrote:
> We have no interest in providing any meaningful level of support for
> Windows or other nonfree operating systems.
> 
This would be useful for less computer literate users as well, I think 
having a simple icon like [0] next to the .. would be a nice help for 
everyone.

[0]: https://fontawesome.com/icons/level-up-alt

Re: Question about authenticating uploads for custom domains on srht.site 7 months ago

From Ignas Kiela to ~sircmpwn/sr.ht-discuss

On 2021-04-18 08:23, demigod@kaction.cc wrote:
> Yet it occurred to me that I do not understand how it works. acurl authenticates user, but I never setup anywhere that ~kaction user is supposed to upload files for kaction.cc website. How it works? What stops me from uploading stuff onto somebody's else custom domain?
> 
It's not configured anywhere, you could technically upload stuff for 
somebody else's domain if they haven't done it yet, but it wouldn't be 
served unless they pointed their domain at pages.sr.ht, at which point 
it would be reported to admins and might bring you some bad reputation 
for that.