<Anarchos>
Begasus: i could flash my bios, and get the wake-on-lan work, but only from within Win7. It doesn't work when shut down from Haiku.
<Begasus>
Hi Anarchos, yeah, just saw your post on the forum
<Anarchos>
Begasus: i wonder if there is a param on the NIC, in addition to the bios
<Begasus>
no idea there Anarchos (as usual) :)
<Begasus>
maybe someone else can step it
<Anarchos>
if i have no answer on the forum i will do a bug report.
<phschafft>
I think I remember that you can set the wol settings in a permanent and a 'next time' way.
<phschafft>
with the idea that if you do the 'next time' way if something goes wrong or the card is swapped things are back to safe defaults.
<phschafft>
but naturally it could also be that Haiku just does some reset of the card on boot that clears the wol settings.
<phschafft>
either way, it sounds like worth a ticket, even if not a bug.
<Anarchos>
phschafft: the rtl8169 driver is a freebsd one. I tried to understand what it does with WOL, but couldn't make my mind
<phschafft>
hm.
<phschafft>
last time I worked with wol was all 3com cards. so some of that might be different with an rtl8169.
<Anarchos>
phschafft: are 3com cards supported on haiku ?
<phschafft>
no odea.
<phschafft>
*idea
<Anarchos>
so you didn't test on haiku :)
<phschafft>
it was before Lenny was released. not sure how long before.
<phschafft>
(Lenny was released in 2009)
<Anarchos>
what is lenny ?
<phschafft>
A Debian release.
<Anarchos>
phschafft: i will make a test to confirm than WOL works for more than one reboot on Win7. Then i will do a ticket.
<phschafft>
ok.
<phschafft>
I can't find a info on my ethtool manpage on that. so I might be wrong or it might be a feature specific to some cards.
<phschafft>
yet it surely is a good test and valueable info for whoever is going to work on that ticket.
<dovsienko>
in my experience, for server purposes WOL got superseded by BMCs and their network interfaces, also by managed power bars
<dovsienko>
for desktops that do not have a management agent built in WOL may still be relevant for massive installation of weekend updates, but my limited practice is that the IT department tells everybody to keep their computers always on
<phschafft>
if I would hear any IT deparntment tell people not to shut down the machine (or hypernate or similar) while not in office I would call them into my office...
<dovsienko>
or office would be crowded if they bothered to show up
<dovsienko>
I understand your reasoning perfectly, but in the real modern world real people forbid engineers to script repetitive commands because it is a waste of time
<dovsienko>
or so they say
<Anarchos>
dovsienko: my use case is to wake up my home server while i am at work, to work through ssh on my real important projects....
<dovsienko>
in the real modern world engineers of "leader market" companies call API functions and do not check if the function call returned an error
<dovsienko>
the "cloud-success-IOT-success-AI-success-success-success" propaganda is just propaganda
<dovsienko>
if a hunder desktops somewhere keeps running overnight and installing occasional software updates, it is a relatively small imperfection
<dovsienko>
Anarchos: ever considered changing the job?
<Anarchos>
dovsienko: no because of the good pay and lack of other software companies in my town
<coolcoder613>
Well, I got a bit further than the last post
<coolcoder613>
I'm reading lines and displaying them
<coolcoder613>
currently working on showing a cursor
<zard>
Is it a bare metal OS?
<coolcoder613>
Oh, and i got backspace working
<coolcoder613>
yes, it's bare metal
<zard>
For x86-64?
<coolcoder613>
cargo run will boot it in qemu
<coolcoder613>
yes
<coolcoder613>
I've run it on real hardware
<coolcoder613>
hardware I *thought* was 32-bit...
<zard>
Interesting... /me looks for the bootloader
<coolcoder613>
github rust-osdev/bootloader
<zard>
Ah. Too bad the bootloader would be the part I would be most familiar with
<zard>
I've written some very simple bootloaders in assembly
<zard>
And if you can get you're program to fit in 512 bytes, you can have the bootloader be the os itself
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
novaphoenix has quit [Quit: gone fishing]
novaphoenix has joined #haiku
* coolcoder613
heads to bed......
<coolcoder613>
g'night
<zard>
good night :)
* Nephele
only just got out of bed
<zard>
Still morning here
<coolcoder613>
time zones ;)
<coolcoder613>
10:12pm thursday night here
<Nephele>
I always have to calculate what people mean when they write am pm
<Nephele>
it's quite confusing
<coolcoder613>
bc your in eu
<coolcoder613>
For me, 24-hour time is confusing
<Nephele>
It's like the US and a couple countries in africa that use 12h time. got little to do with the EU
<coolcoder613>
I am honest enough with myself to admit it's the better system
<coolcoder613>
I'm in AU
<coolcoder613>
We use 12-hour here
<Nephele>
wikipedia sais "both in common use" for AU, but not sure how accurate that is, of course
<coolcoder613>
AFAIR I've never seen it used
<Nephele>
as for better system, maybe it's slightly better? but then the time system in general is a bit icky
<coolcoder613>
Possibly in different regions fo AU
<coolcoder613>
*of
<Nephele>
so not sure what would be gained by moving to a slightly better system for you :P
<Nephele>
anyway, you wanted to sleep. So have a good night :)
<coolcoder613>
nephele: It's superior in the same way metric is over imperial, and celsius is over farenheit
<Nephele>
I don't see how celsius is superior to fahrenheit. It is equivalent (stepwise) to Kelvin which is used by the metric system, so in that way it is important. but other than that, eh.
<coolcoder613>
nephele: I'm not talking about moving to it, I just think it's a better system if I had started with it in the first place
<Nephele>
If there was a better time system with base10 for example you could avoid all those nasty * 3,6 conversions in metric
<Nephele>
So? is defining temperature based on some random molecule mix boiling and freezing at some specific pressure less arbitrary?
<coolcoder613>
Practically, yes
<Nephele>
I've heard arguments for the fahrenheit scale that say that the 0-100 range there more closly matches what humans are generally living in
<Nephele>
and i also can see the logic in that
<coolcoder613>
but mainly, it boils down to me being used to celsius
<coolcoder613>
and not seeing farenheit as any better
<Nephele>
sure
<coolcoder613>
and getting annoyed when I see measurements in farenheit
<coolcoder613>
which makes me dislike it
* coolcoder613
really heads to bed
<Nephele>
Heh, i can understand that. it's more of a cognitive bias though. I don't *understand* what the numbers in fahrenheit are supposed to mean
<Nephele>
nighty night
<phschafft>
nachti coolcoder613!
<Anarchos>
good night coolcoder613
<Nephele>
Begasus: around?
<dovsienko>
when the user disables virtual memory, is there any need to keep /boot/system/var/swap?
<dovsienko>
on an unrelated note, is "df" supposed to display the actual amount of space used in RAMFS instead of 0.0?
tuaris has joined #haiku
<Nephele>
Is there any benefit to letting users disable swap? seems wierd to me to even have that visible.
<Anarchos>
dovsienko: you can delete the swap file when virtual memory is disabled
<Anarchos>
nephele: it is easier to test low (memory) resource without virtual memory.
<dovsienko>
Anarchos: do you think it would make sense to delete it automatically?
<phschafft>
I have many systems without swap. and given the nature of SSDs I would generally consider to be on-disk swap as an opt-in thing in 2024.
<Nephele>
disabeling swap means the virtual memory subsystem can't evict RW pages, only RO pages that came from the disk
<phschafft>
or pages marked as such.
<dovsienko>
depends on the use case and what the host is doing on a typical day/week. in my use case it is more effective to beef the VM up than to have it swapping
<waddlesplash>
yes, I also disabled swap, since I'm on SSDs
<Nephele>
ah, well a VM may be a different case alltogether
<waddlesplash>
if the file isn't deleted automatically that sounds like a bug yes
<waddlesplash>
and ramfs should account for total size but doesn't yet, we should fix
<dovsienko>
let me file an ticket for that
<phschafft>
e.g. POSIX allows for a process to deallocate but not unmap pages.
DKnoto has quit [Ping timeout: 480 seconds]
<dovsienko>
#18971 and #18972 it is
<dovsienko>
upon a closer examination it turns out Haiku df does not have the "used" size, so in this case it was the "total" that is zero
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
<Nephele>
kallisti5[m]: I don't have that issue (57935)
<Nephele>
i did have a similar issue when i had a broken disk though
<kallisti5[m]>
nephele: which input device?
<kallisti5[m]>
usb mouse on real hardware, vm, etc?
<Nephele>
apple might mouse on usb hub (apple aluminim keyboard) connected over xhci controller to a dell optiplex, no vm
<waddlesplash>
kallisti5[m]: weird, haven't heard that reported by anyone. that commit should only affect things when apps use custom cursors. using regular cursors that code should never be hit at all
<kallisti5[m]>
trying to reproduce in a vm now to see if I can reproduce it anywhere else
<Nephele>
your issue sounds like something in input_server went awry
<kallisti5[m]>
yeah, I need to check syslog still. I had the issue across multiple reboots. It might be something weird on my end. (non-packaged, etc) checking
<Nephele>
booting with user-addons disabled might rule that out
<waddlesplash>
ohh
<waddlesplash>
kallisti5[m]: it's probably your mouse settings
<waddlesplash>
grab a copy of the settings file and upload it please
<waddlesplash>
I bet deleting it will fix the problem
<kallisti5[m]>
no input-server stuff in non-packaged... but I do have an xhci in non-packaged/add-ons/kernel/busses/usb/
<waddlesplash>
that likely won't even be loaded honestly
<waddlesplash>
XHCI gets loaded by the bootloader from a specific directory
<waddlesplash>
don't think it checks non packaged
<kallisti5[m]>
added Mouse_settings to ticket
<kallisti5[m]>
plopping it into a vm to see if I can reproduce it
<waddlesplash>
input server may overwrite it on shutdown, I can't remember
<kallisti5[m]>
nah, last modified 2023
<waddlesplash>
ah. ok
<kallisti5[m]>
doesn't break qemu, but the settings file seems to specify "USB Mouse 1" while qemu is PS/21
<kallisti5[m]>
s/21/2/
<Nephele>
you can specify to use a usb mouse in qemu if you like
<kallisti5[m]>
yeah. just read that -device usb-mouse
<kallisti5[m]>
confirmed. reproduced
<kallisti5[m]>
confirmed erasing that Mouse_settings, running a sync, and rebooting resolves the issue
<kallisti5[m]>
added notes to ticket
<waddlesplash>
should be an easy fix
<waddlesplash>
madmax will hopefully come along with a patch in short order
<waddlesplash>
but I think we can branch the release today and just deal with this in later builds
<kallisti5[m]>
yeah, I do think it is a blocker because i'm expecting it to impact R1/beta4 -> R1/beta5 upgrades
<waddlesplash>
agreed
<kallisti5[m]>
do we want to try and use libglvnd on r1/beta5 ?
<waddlesplash>
if it works well enough
<kallisti5[m]>
it's "better", but SDL bugs still exist
<waddlesplash>
then maybe not
<kallisti5[m]>
hm. It'
<kallisti5[m]>
ok
<waddlesplash>
let's get those resolved. keep in mind we can switch to glvnd between releases
<kallisti5[m]>
yeah.. it's just a hard upgrade
<waddlesplash>
we don't need to wait for the next beta or whatever to start using it, we can switch to it in the nightlies and for existing beta users
<kallisti5[m]>
also, once we build for nightly, stable releases will upgrade too
<waddlesplash>
yes
<waddlesplash>
but I don't think that's an issue
<waddlesplash>
but also I want to maybe work on HaikuPorter after the release, so maybe we could get separate build streams for HaikuPorts
<kallisti5[m]>
true. ok We can put off libglvnd until after reviewers review R1/beta5 lol
<waddlesplash>
but that's a whole other thing
<kallisti5[m]>
yeah, we're really getting limited by haikuporter
<kallisti5[m]>
(on the infra side)
<waddlesplash>
yeah
<Nephele>
kallisti5[m]: reviewers will do the same thing as last year :D "Why does my youtube channel not work? bad!"
<phschafft>
is there a list of all filesystems Haiku can mount rw?
<waddlesplash>
don't think so? maybe on the wiki
<Nephele>
not that many :g
<waddlesplash>
off the top of my head: bfs (duh), fat, ntfs, ext2-4
<waddlesplash>
some write code for btrfs but it's incomplete and disabled
<Nephele>
ufs2? or is that RO
<waddlesplash>
I think there's at least 1-2 more. does our UFS driver support rw?
<phschafft>
ext* is high on my compatibility list. so if ext4 writing works without problems that may already solve my problem.
<waddlesplash>
check open tickets
<waddlesplash>
I think there may be a bug or two remaining for ext4
<waddlesplash>
but the code is very good so it should not be hard to debug
* phschafft
nods.
<zard>
Yeah, I think I had some issues with putting WebKit's source code onto ext4
<phschafft>
plus there is always a difference between bugs that corrupt stuff and bugs that just mean some optional features don't work or something.
<phschafft>
btrfs would also be nice.
<dovsienko>
phschafft: RAMFS :-D
<phschafft>
dovsienko: I mean at least that will not have problems with on-disk corruption! ;)
<Nephele>
if you write your driver creatively anything is possible
<dovsienko>
it is getting better, one bug at a time
<Nephele>
:)
<phschafft>
if ext4 is very far it could also be used as a primary filesystem for the system. that would help e.g. when running in a VM. allowing easy access from both the guest and the host.
<kallisti5[m]>
please never look at my buildmaster commits... they're horrible :P
<Nephele>
ext4 can't be used for haiku as primary FS aslong as the extended attributes are still size limited
<dovsienko>
no OS is perfect. I remember discovering that OpenBSD steadily OOM-crashes on AArch64 with gigabytes of RAM still free, if and only if there is no swap configured
<dovsienko>
immediately after reporting this to OpenBSD developers they began to maintain dead dead dead silence as if I offended everyone personally
<Nephele>
If it's not perfect then patches are welcome ;)
<kallisti5[m]>
⚠️ NOTICE: Haikuports package builders have resumed functionality. Hold off on merging anything haikuports though for a little while. waddlesplash prioritizing a few builds of things.
<Anarchos>
oh beta5 on the way ?
<Nephele>
yes, we branched
<Anarchos>
nice news !
<kallisti5[m]>
now all that's left is "everything else"
<Nephele>
Heh :)
<Anarchos>
kallisti5[m] nespresso ?
<kallisti5[m]>
I have a machine somewhere, but it needs cleaned out