This space intentionally left blank.


Last active 12 hours ago


Last active a day ago


Last active 2 days ago


Last active 9 days ago
View more

Recent activity

seatd 0.4.0 2 days ago

From Kenny Levinsen to ~kennylevinsen/seatd-announce

Announcing the release of seatd 0.4.0.



seatd 0.4.0

Notable improvements in this release includes several fixes to IPC
connection handling, VT switching and FreeBSD portability.

Kenny Levinsen (23):
      poller: Raise signals through self-pipe
      client: Remove unnecessary active_client check

Re: Do you plan to tag a release for wlsunset? 6 days ago

From Kenny Levinsen to ~kennylevinsen/public-inbox

October 15, 2020 6:43 PM, "Drew DeVault" <sir@cmpwn.com> wrote:

> I would like to package this for Alpine Linux but I'd prefer to wait
> until a release is tagged.

Currently stuck trying to handle the arctic edge-cases of midnight sun/polar
night in a smart way, which currently mess up my solar elevation angle math.

I'll probably tag a release after I find a solution for that. I'll ping you
when it happens.

Re: seatd fails to build with GCC on FreeBSD due to -Werror a month ago

From to ~kennylevinsen/public-inbox

Included in 0.3.0.

August 30, 2020 8:19 PM, "Jan Beich" <jbeich@freebsd.org> wrote:

> FreeBSD < 13 on powerpc*, mips*, riscv64, sparc64 still use GCC by default.
> $ pkg install gcc10
> $ CC=gcc10 meson _build && meson compile -C _build
> [...]
> ../seatd/client.c: In function 'get_peer':
> ../seatd/client.c:45:51: error: operand of '?:' changes signedness from 'int' to 'gid_t' {aka
> 'unsigned int'} due to unsignedness of other operand [-Werror=sign-compare]
> 45 | *gid = cred.cr_ngroups > 0 ? cred.cr_groups[0] : -1;
> | ^~

Re: [PATCH seatd v2 0/4] Add support for custom log callbacks a month ago

From Kenny Levinsen to ~kennylevinsen/public-inbox

August 27, 2020 5:56 PM, "Simon Ser" <contact@emersion.fr> wrote:

> This implements the changes we discussed in the first review.
> Simon Ser (4):
> Rename enum log_level to libseat_log_level
> Introduce libseat_set_log_handler
> Introduce libseat_set_log_level
> common/log.c | 74 +++++++++++++++++++++++++++++------------------
> include/libseat.h | 33 +++++++++++++++++++++
> include/log.h | 26 +++++++----------
> libseat/libseat.c | 13 +--------

Re: [PATCH seatd 2/2] Introduce libseat_set_log_handler a month ago

From Kenny Levinsen to ~kennylevinsen/public-inbox

August 27, 2020 1:57 PM, "Simon Ser" <contact@emersion.fr> wrote:

> Yes, you're right, we'll need to add a NULL check in log_init too.
> I had the NULL check in place in libseat_set_log_handler to allow callers to
> restore the default log handler. Hard to tell without docs, I'll add them in
> the next patch.


Will this include amending the loglevel enum changes to be log_level?

> Both libinput and wlroots:

Re: [PATCH seatd 2/2] Introduce libseat_set_log_handler a month ago

From Kenny Levinsen to ~kennylevinsen/public-inbox

August 26, 2020 11:22 AM, "Simon Ser" <contact@emersion.fr> wrote:

> +void log_init(enum libseat_loglevel level) {
> + if (start_time.tv_sec >= 0) {
> + return;
> + }
> + clock_gettime(CLOCK_MONOTONIC, &start_time);
> + current_log_level = level;
> + current_log_handler = log_stderr;
> + colored = isatty(STDERR_FILENO);
> +}

log_init is called in libseat_open_seat, which will presumably be after the user set their preferred log handler.

Re: [PATCH seatd 1/2] Rename enum log_level to libseat_loglevel a month ago

From Kenny Levinsen to ~kennylevinsen/public-inbox

August 26, 2020 11:22 AM, "Simon Ser" <contact@emersion.fr> wrote:

> This is preparatory work for exposing a public function to set libseat's
> log handler.
> No space between "log" and "level" for consistency with the enum
> members.

Now you mention it, maybe there should be a space. Might just have been a Danish compound word leak on my end. What do you think?

[PATCH soju] service: Handle zero-value in stringPtrFlag.String 4 months ago

From Kenny Levinsen to ~emersion/public-inbox

FlagSet.PrintDefaults uses reflection to construct a zero value, calls
.String on it, and compares the result with the current flag value to
detect zero-value flags. For stringPtrFlag, this would result in a
panic, as String() always dereferenced the first level of its **string.

Add another check so that both pointer levels are nil-checked.
 service.go | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/service.go b/service.go
index ef708e1..85dfe19 100644
--- a/service.go
+++ b/service.go
[message trimmed]