~emersion

France

https://emersion.fr

I'm Simon Ser, I write open-source software.

~emersion/goguma-dev

Last active a day ago

~emersion/soju-dev

Last active 2 days ago

~emersion/public-inbox

Last active 2 days ago

~emersion/grim-dev

Last active 6 days ago

~emersion/imap-protocol

Last active 3 months ago

~emersion/mrsh-dev

Last active 11 months ago

~emersion/gqlclient-dev

Last active 1 year, 7 days ago

~emersion/alps-dev

Last active 3 years ago

~emersion/drm-constraints

Last active 3 years ago
View more

Recent activity

Re: [PATCH v2] auth/pam: allow specifying PAM service name as driver param 2 days ago

From Simon Ser to ~emersion/soju-dev

Yeah, I think always using "soju" would make more sense (with a default
PAM config file which has "auth include login", like swaylock).

I don't think this would be a breaking change, since the behavior with
a newer soju (with the PAM config file) would be the same?

One annoying detail that we'd need to handle is installing the PAM config
file conditionally if PAM is enabled at build time… Maybe it's time to
introduce a Makefile configuration mechanism or something?

Re: [PATCH kimchi] Allow 'site' directive to listen on a single IP address 2 days ago

From Simon Ser to ~emersion/public-inbox

Ah. To be honest, this was by design: the host used for matching requests
(the Host HTTP header field) is different from the host used for listening.

Related feature request in tlstunnel: https://todo.sr.ht/~emersion/tlstunnel/26

Re: [PATCH] Don't scale single pixel selection 6 days ago

From Simon Ser to ~emersion/grim-dev

On Saturday, May 18th, 2024 at 13:38, Thomas Weißschuh <thomas@t-8ch.de> wrote:

> When the user requests a single pixel, ignore the output scaling, as
> they probably really only want a single pixel.
> 
> The colorchooser in xdg-desktop-portal-wlr requests a single pixel from
> grim. It then validates that it only received a single pixel in
> exec_color_picker().
> On scaled displays grim would return more pixels than expected and
> the color picking would fail.

I think it would make more sense to explicitly specify -s in
xdg-desktop-portal-wlr.

Re: [PATCH] fix(api)!: change default config filename on darwin systems 6 days ago

From Simon Ser to ~xenrox/hut-dev

I don't think this is a good idea. os.UserConfigDir returns the standard
directory for the OS, XDG_CONFIG_HOME is not a standard on macOS, it's
just abused by software coming from the FreeDesktop.Org world. Having
special macOS-specific code to re-implement the FreeDesktop.Org standard
on a platform it's not supposed to be used for is incorrect.

Re: Re[2]: [PATCH 2/3] seatd: use systemd socket activation 9 days ago

From Simon Ser to ~kennylevinsen/seatd-devel

On Wednesday, May 15th, 2024 at 12:18, Anthony <burning2007@ya.ru> wrote:

> Please, stop forcing marketing bullshit to good software

This comment is not appropriate. Not only you're using derogatory
language, but also the comment doesn't add anything to the discussion.
This is not the place to vent about systemd.

Re: [PATCH 2/3] seatd: use systemd socket activation 9 days ago

From Simon Ser to ~kennylevinsen/seatd-devel

On Wednesday, May 15th, 2024 at 10:05, Michael Tretter <m.tretter@pengutronix.de> wrote:

> To me, re-implementing/copying the sd_listen_fds() function in seatd
> seems like a really bad compromise for implementing socket activation
> without going full systemd.

Note, the trend lately has been to *not* link with libsystemd for small
helpers (see xz/openssh). The systemd docs now even have code snippets
to copy/paste for basic helpers.

Re: [PATCH] pager: completely write output when not on terminal 10 days ago

From Simon Ser to ~xenrox/hut-dev

On Monday, May 13th, 2024 at 21:01, Jens Schmidt <farblos@vodafonemail.de> wrote:

> 1. Would you accept a patch explaining the background in the man
> page?

Not sure how any of this is relevant to end-users TBH.

> 2. And what about providing automatic cursor management in command
> graphql? Would you reject that for the same (DoS) reason?

Why not. Not super sure how that'd work though.

> 3. After all, to some extent this is security by obscurity, isn't
> it? If I script some automatic cursor management around graphql,

Re: [PATCH] pager: completely write output when not on terminal 11 days ago

From Simon Ser to ~xenrox/hut-dev

We don't want to DoS the server, so no, can't make this optional.

Re: [PATCH wlr-randr] add a new special output name `*` 14 days ago

From Simon Ser to ~emersion/public-inbox

Sorry for the delay.

Can you expand on your use-case of this? Disabling all outputs is not
the same as powering off all outputs: disabling an output moves away
windows and UI elements, powering off an output leaves them as-is.