egrath has quit [Quit: Vision[]: i've been blurred!]
egrath has joined #haiku
LinuxUser has quit [Remote host closed the connection]
LinuxUser has joined #haiku
LinuxUser has quit [Remote host closed the connection]
LinuxUser has joined #haiku
matt3 has joined #haiku
AlaskanEmily has quit [Remote host closed the connection]
matt3 has left #haiku [#haiku]
LinuxUser has quit [Remote host closed the connection]
LinuxUser has joined #haiku
OscarL has joined #haiku
<OscarL>
Good morning netpositive, andreasdr[m], Begasus.
* OscarL
should pay more attention to the logs' timestamps :-D
<Begasus>
Hi OscarL :)
* OscarL
waves hand at Begasus, while still half-asleeep :-P
<Begasus>
what's the time there?
<OscarL>
7:30 AM.
<Begasus>
ah, early start then, 11:30AM here
<Begasus>
k, strike libfec :)
<OscarL>
still finding static libs to remove?
<Begasus>
nah, just a side track ;)
<Begasus>
prepared rtl_sdr recipe here for future PR when it arives ;)
* OscarL
wishes he had a compatible receiver to play with rtl_sdr :-)
rennj has quit [Ping timeout: 480 seconds]
<OscarL>
sounds ultra fun.
<Begasus>
don't have the hardware, but can build it :)
<OscarL>
meanwhile... I'm attempting a bit of light maintenance on some of the acpi drivers under "kernel/driver/power/".
<OscarL>
but can't seem to be able to "reload" them without an OS restart (which is a PITA in the slow netbook I'm doing that).
<Begasus>
nice
<OscarL>
So... I'm farily sure that just removing/replacing the binary for a driver triggers a reload, at least for some drivers... maybe only the ones using the "legacy" driver interface?
<OscarL>
Does anyone knows if it is possible to reliably "reload" a new version of a given binary driver using the new driver API? Should that work already? Doesn't works in my case (say... with "acpi_ac") for some ACPI reason, or something?
<Begasus>
No idea there ;)
<OscarL>
I can, of course, just reboot.... what kills me is that doing so, "git" then has to "cold-start" again even for the most basic things (nothing on cache memory), and git I/O on large repos is SLOOOOOW on that Atom N450 netbook :-(
rennj has joined #haiku
<OscarL>
With the help of the kind reviewers, managed to land a patch already for "acpi_ac" (and a small build fix for "acpi_lid"). I'm happy :-)
<Begasus>
;)
<OscarL>
Already eyeing another small one for "acpi_battery".... Autocommiter will not stay above me for long on the stats! :-P
<Begasus>
go for it! :)
<Begasus>
Need some sick leave to get back on the sadle :P
<OscarL>
too much "real" work Begasus?
<OscarL>
or just the usual "not enough hours in a day"?
<Begasus>
yeah, when I'm at work and combined with the dogs I don't get around to doing much work on the laptop :)
<OscarL>
We appreciate (and benefit from) any and all amount of time you put into Haiku-related things, Begasus. Much appreciated!!!
<Begasus>
thanks :) I'm not alone you know :)
<OscarL>
So.... :-).... noticed that "acpi_thermal" driver (not included on the oficial images/packages) basically works, but reports temps on Kelvin degrees :-/
<Begasus>
don't know that, but maybe would scare some people not familiar with it ;)
<OscarL>
372.2 degreees, yeah.... this thing is HOT :-D
<OscarL>
"pch_thermal" uses Celsius when you "cat /dev/power/pch_thermal" (at least that's what it seem, don't have hardware to test it), so I'm planning to make acpi_thermal do the same.
<OscarL>
also, managed to read the "core-temp" from my Phenom II CPU via pci config registers, so a possible "amd_temp" driver (similar to what pch_thermal does for Intel CPUs) might be in my future.
<OscarL>
but having "acpi_thermal" working, and hopefully being able to show its reading from ActivityMonitor (and/or a deskbar icon) would be a good start.
<Begasus>
meanwhile, libcorrect base recipe done :)
<OscarL>
to quote andreasdr[m]....: nice!
<Begasus>
;)
* OscarL
likes that andreasdr[m] is always encouraging people on their efforts.
<OscarL>
Begasus: does rtl_sdr has lots of dependencies?
<OscarL>
s/does/do/, s/has/have/ (I think)
<Begasus>
nope, only libusb
<Begasus>
whell from my attempt to build it :)
<Begasus>
heading out, cu later!
<OscarL>
Later Begasus! be well!
LinuxUser has quit [Remote host closed the connection]
<OscarL>
it forced the OS to rescan the name driver (if given one), or to rescan for changes in drivers.
<OscarL>
s/name/named/
<OscarL>
testing changes in "acpi_thermal" while having to reboot to get the new code to load up is no fun :-D
<jessicah>
oh I see
<OscarL>
I've got another WIP driver (for IT87xx LPIO chips) that uses the legacy interface, an that one does reloads automatically when removed/re-copied under "~/config/non-packaged/add-ons/kernel/[...]". No such luck with the acpi_* ones, sadly.
Skipp_OSX has quit [Ping timeout: 480 seconds]
<OscarL>
Wonder if the difference is due to the newer driver/module API... due to some ACPI thing..., or just my usual bad luck, and I just hit some bug.
<OscarL>
deleting, say ""~/config/non-packaged/add-ons/kernel/power/acpi_thermal" makes "cat /dev/power/acpi_thermal/0" fail with "No such file or directory" (albeit "ls /dev/power/" and "ls /dev/power/acpi_thermal/" still show that dir and file).
<OscarL>
copying a modified "acpi_thermal" binary on the above "non-packaged" dir... makes the "cat" call work again... but with the behavior of the previous binary.
DrachenMaus has joined #haiku
<OscarL>
Not sure if bug or feature, basically :-D
DragonMaus has quit [Read error: Connection reset by peer]
DrachenMaus has quit [Read error: Connection reset by peer]
DragonMaus has joined #haiku
<jessicah>
I would classify as a bug :p
<jessicah>
or as, needs implementation
<nephele>
OscarL: did you blacklist acpi_thermal?
<nephele>
If yes then i would expect that behaviour
<OscarL>
nephele: it is not part of the images yet, so no need to blacklist this particular one.
<nephele>
well, if it disapears i would expect it's device nodes to disapear
<nephele>
or i guess you mean the "behaviour of the previous binary" as a bug?
DrachenMaus has joined #haiku
<OscarL>
(same issue with "acpi_ac" and "acpi_lid", none of them part of beta/nightlies yet)
<nephele>
well, if it really behaves like the old image that does sound like a bug definetely
<OscarL>
nephele: yeah... if I replace the binary... I would expect it to use that instead of the one that seems to keep loaded somewhere :-D
<nephele>
yeah, that sounds like a big headache!
<OscarL>
specially because I'm dumb... so it took me way too much work to figure that was (one of) the problems :-D
<nephele>
If the system behaves unexpectandly that makes debugging much harder, definetely file a ticket i'd say
<nephele>
or even better would be fixing it :D
<OscarL>
ha! fat chance if that's on me :-P
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
<OscarL>
Thanks jessicah, nephele. I guess I'll file a ticket then.
DragonMaus has quit [Ping timeout: 480 seconds]
<tqh>
IIRC, the new driver api does not implement reload yet.
<OscarL>
tqh: Thanks. That would explain the difference I've seen with some of my old drivers.
<tqh>
yes I started using the new API long ago and ran into that myself.
<OscarL>
Too bad, thou, because each reboot causes calls to git on that netbook to take forever, as nothing is in the I/O cache... > 5 mins for "git status" :-(
<tqh>
It seems that the git status slowness would be good to fix as well
<nephele>
Yeah, it's one of the more annoying parts about developing on haiku :)
<OscarL>
while I have you here... do "git push" to gerrit also takes between 3-7 minutes in the "remote: Counting objects" step of the push for you too?
<OscarL>
I've seen that since my first patch in August 2022.
* OscarL
wonders if gerrit is in need of some "git gc"
<nephele>
I don't remember honestly, but i also rebase my stuff before pushing it to gerrit, maybe that helps?
<nephele>
I guess that might be a good question for the mailing list
<nephele>
I feel like making developer tools faster is a good way to get more work done... figuring out why git is so slow would be a massive speedup, it shouldn't be this slow on my ssd :(
<OscarL>
I'm always up-to-date when I push. The patch upload is fast... only that "remote: Counting objects xxxxx". And the number grows ever larger, of course :-D
<tqh>
one trick for more speed is to use ed25519 for git ssh instead of rsa.
<nephele>
huh, that may be it. rsa didn't work at all for me
<nephele>
OscarL: Yeah, would be interesting to ask on the ML, other devs can chime in then too... perhaps it's a bit of a bigger problem
<OscarL>
I'm using "id_ed25519".
<tqh>
then you can't get improve that way..
<OscarL>
nephele: cool... I'l have to dust up my mail address (barely touch email since I left work :-D)
<nephele>
don't have your own cool email server? :D it's really suprising how bad mail server software is!
<nephele>
postfix keeps suprising me with yet-another-thing-that-clearly-should-have-been-default options beeing set to whatever was cool in the 90s i suppose
<nephele>
I should really sit down and re-do my mail setup with opensmpt... they seem to be all about that stuff (beeing an openbsd project)
<nephele>
opensmtpd
<nephele>
anyhow, got of track here :)
<jessicah>
I haven't had too many issues pushing
<OscarL>
I used to love email in the Mail Daemon Replacement days... then I gmail started to lock me out of most email clients... until I couldn't bother any longer.
<tqh>
perhaps network driver issue?
<jessicah>
I was running my own mail server for a while, but then I moved and can't get fibre any more, nor a static ip
<OscarL>
the weird thing is the slowest step for me is a "remote:" one... as if I'm being throtled down, or sent to a low prio queue :-D
<tqh>
perhaps fragmented packets. test with lower MTU?
<OscarL>
tqh: tried from bare-metal on 2 different machines, from VMs in VirtualBox, VMware, Qemu+KVM... all the same :-(
<nephele>
jessicah: not even static ipv6?
<tqh>
OscarL, all haiku vms?
<nephele>
For my part i'm renting a vps for my mail server, that's not really 100% "secure" but it certainly beats giving emails to google directly .-.
<OscarL>
tqh: yes.
<tqh>
And fine outside Haiku?
<OscarL>
Can't tell, as I haven't pushed to gerrit from anything but Haiku.
<OscarL>
pusing to github, even from big repos (my Haiku clone), work inmediatly.
<jessicah>
nephele: well, really, the biggest issue is getting ISP to configure reverse-DNS
<nephele>
I don't really get why reverse DNS is that relevant for mail
<tqh>
My guess would be it is something with Haiku IP stack or network conf. Lowering max packet size in Haiku or comparing packet size with other OS might help.
<nephele>
I suppose to figure out which mail server is connecting to you? hmm
<OscarL>
tqh: thanks, will keep that in mind.
<nephele>
though i don't know if smtp doesn't already include that info
<jessicah>
well, can get static ip, but it's like $10/month, which is ridiculous
<nephele>
jeez .-.
<nephele>
I don't get a static ipv4, but i do get a static ipv6 so that would be feasible anyhow
<nephele>
though... haiku wouldn't be able to connect to my mail server then, which would suck :D
<jessicah>
nephele: mail servers use it to validate that they're connecting to a machine matching the DNS
<jessicah>
it's part of spam/abuse detection
<nephele>
yes, i know that part
<nephele>
but why do you need reverse dns fo it?
<nephele>
I assume you are talking about spf, and dkim
<jessicah>
reverse dns is how you go from ip to dns name
<nephele>
Yeah. doesn't smpt already provide the dns name though?
<nephele>
ugh, i'm not looking forward to reading up on all the mail security things again :D
<nephele>
Maybe I can add mail signing with s/MIME this time around though
<jessicah>
:)
<nephele>
err, encryption, not signing
<jessicah>
I just know that I needed it for a few mail servers
<jessicah>
particularly in EU
<nephele>
Yeah, I do believe that. It is correctly configured for my vps so it likely was never an issue
<jessicah>
as they do a PTR lookup on IP, and then compare to allowed hostnames in your SPF DNS record
<nephele>
just curious if smtp even needs it, but i suppose that is really moot if servers don't use that info
<jessicah>
SMTP itself doesn't, it's the layers on top
<nephele>
Yeah, i've always had "don't care" in my spf record
<nephele>
I figure if my domain is cryptographically signed with DKIM to come from me you should trust that
<nephele>
don't really see the advantage of the check of source name other than breaking mailing lists
<jessicah>
even with that in SPF, some servers will use the lack of explicit PTR to identify as a "home" server
<nephele>
err if the envelope is signed with DKIM i mean
<jessicah>
OscarL: just did a push to review on linux, only took a handful of seconds
<OscarL>
thanks jessicah. Then it is my usual bad luck :-D
<jessicah>
^ that will also fix your zstd_devel missing
<OscarL>
nice! :-)
<OscarL>
Another question... do package downloads / Haiku updates tops around 230/240 KB/s for everyone else, or that's also just me?
<jessicah>
I sometimes see that within haiku, yeah
<jessicah>
sometimes a bit more, it's a little variable, but often slow
<OscarL>
mmm using "wget" to download some random iso... same cap... but yt-dlp... downloads above 1.5 MB/s while the wget is still running at 240 KB/s :-D
<OscarL>
(ActivityMonitor validates the speeds reported by those tools)
HaikuFan has joined #haiku
DrachenMaus has quit [Read error: Connection reset by peer]
DragonMaus has joined #haiku
HaikuFan has quit []
gouchi has joined #haiku
<nephele>
Hmm, mouse won't move anymore
<nephele>
and if i open input preferences it's .... just blank
<OscarL>
sweet jebuz.... "time git commit --amend"... real 4m16.062s. (minus the 2 seconds it took me to close the editor)... that's why I would love to avoid having to reboot to test driver changes :-D
jmairboeck has quit [Quit: Konversation terminated!]
<OscarL>
At least now "cat /dev/power/acpi_thermal/0" shows: "Current Temperature: 60.0 ºC" instead of that (correct but weird looking): "333.2 K" :-D
<jessicah>
OscarL: you could probably try setting commit.status to false in your config
<jessicah>
then you shouldn't have to wait for the stat madness
<jessicah>
kelvin? hah
<OscarL>
(re: git) kewl! Will certainly try that jessicah.
<OscarL>
Yeah... ACPI's Thermal Zones uses deci kelvins.
OscarL has quit [Remote host closed the connection]
OscarL has joined #haiku
nephele has quit [Quit: Vision[]: i've been blurred!]
gouchi has quit [Remote host closed the connection]
<OscarL>
Aparently some implementations use 2731 dK (while others use 2732 dK) to mean 0 Celsius (instead of the exact 273.15 K)
HaikuUser has joined #haiku
<OscarL>
jessicah: thanks a bunch for that "commit.status=false" tip! my decade+ hardware certainly needs it :-D (at least for the Haiku an HaikuPorts repos)
<jessicah>
:)
HaikuUser2 has joined #haiku
B2IA has quit [Quit: Vision[]: i've been blurred!]
rennj has quit [Ping timeout: 480 seconds]
B2IA has joined #haiku
HaikuUser2 has quit [Quit: Vision[0.10.3]: i've been blurred!]
HaikuUser has quit [Quit: Vision[0.10.3]: i've been blurred!]
tqh has quit [Quit: Leaving]
rennj has joined #haiku
egrath has quit [Ping timeout: 480 seconds]
rennj has quit [Ping timeout: 480 seconds]
hightower2 has joined #haiku
rennj has joined #haiku
rennj has quit [Ping timeout: 480 seconds]
rennj has joined #haiku
DrachenMaus has joined #haiku
DragonMaus has quit [Read error: Connection reset by peer]
DrachenMaus has quit [Read error: Connection reset by peer]
DragonMaus has joined #haiku
_Dario_ has quit [Quit: Vision[]: i've been blurred!]