<nekobot>
[haiku/haiku] 74be257edff7 - udp: send should return the error on the last send packet
<nekobot>
[haiku/haiku] 48e3aa808107 - kernel/socket: allow sendto to an empty address
<nekobot>
[haiku/haiku] 2b9fdf784780 - ipv4: empty address should be a zero sa_len
<Begasus[m]>
tinkering about setting up some build for master branches Qt (should be 6.10.0) :)
<OscarL>
good to have things going smooth from time to time. sometimes things can get "too interesting" otherwise :-)
<Begasus[m]>
right :)
<OscarL>
Haiku detects my newly ~~stolen~~ acquired Bluetooth USB dongle as "TP-LINK UB500 Adapter <?><?><?><?><?><?><?><?>"
<OscarL>
someone didn't null-ended a string somewhere.
Nasina has joined #haiku
<Begasus[m]>
got a bug solved upstream after reporting it for kirigami, those dev's still are amazed how well things look on Haiku :)
Nasina has quit [Read error: Connection reset by peer]
<OscarL>
Begasus[m]: kinda funny how some KDE things look way more at home on Haiku than they do on Linux :-)
<Begasus[m]>
I'm glad the haikuplugin is working fine so far on Qt6.9.* couldn't get that right on 6.8.*
<OscarL>
(Bluetooth preference needs some love from someone good with Layout GUI... where's humdinger when you need him? :-P)
<Begasus[m]>
not around :P
<OscarL>
Seems I'm not able to pair Haiku with the phone (apparently we can't do much more than that anyway).
<OscarL>
I would like to have Bluetooth PAN working.
<Begasus[m]>
probably needs more work in the base before even tackling the UI
<Begasus[m]>
patches welcome? at least you got something to check with :P
<OscarL>
(Also Bluetooth HID would be nice, to be able to hook it up to something like KeyCursor... but using some old phone as mouse/touchpad)
<Begasus[m]>
ah, so you want to swipe to move to a different workspace? :)
<OscarL>
re: patches... I need to read a lot about this before even knowing if it's something anywhere near my skill level :-)
cassisian has joined #haiku
<OscarL>
haven't really even used bluetooth more than on a couple of occasions, and just for quick tests.
<OscarL>
so I have a lot to catch up :-D
Nasina has joined #haiku
<Begasus[m]>
keeps you on edge :)
cassisian has quit [Ping timeout: 480 seconds]
OscarL-64 has joined #haiku
OscarL has quit [Ping timeout: 480 seconds]
OscarL-64 is now known as OscarL
<OscarL>
vision is eating part of one core... in thread "s->chocoak_is_my_hero"
* OscarL
will "debug" it to see where's that... laterz!
<OscarL>
mmm, probably still trying to read from an old socket or something (had a network disconnect-reconnect here)
OscarL has quit [Quit: Gone with the wind]
OscarL has joined #haiku
<Begasus[m]>
so far no ghost threads here :)
<OscarL>
k. Managed to boot Haiku again with the Phenom, updated haiku/haikuports repos... will try to tackle Python 3.13.5 test suite again tomorrow.
Aedil has quit [Ping timeout: 480 seconds]
<Begasus[m]>
nice!! crossing fingers for you!
<OscarL>
Old motherboard has some capacitors that get so hot that burned one of my fingers a couple of days ago. Will try to re-cap it, but something else has to be really wrong for those caps to get so hot.
<OscarL>
(and I don't have neither the tools, nor the brain, to properly diagnose it :-/)
<OscarL>
Thanks Begasus[m]. Haiku feels really snappy on this Phenom while on bare-metal. Hoping that damn test-suite gives me less problems here than on the netbook.
<qth[m]>
Managed to get my network working with ipv4 :)
<qth[m]>
Where can I put commands that will be executed on boot?
<OscarL>
there's an "UserSetupEnvironment.sample" in that dir, just in case.
<Begasus[m]>
k, biab :)
<OscarL>
not sure if default install has a "UserBootscript.sample" too, this install has some age already.
<qth[m]>
I have one more general question: I often stumble on command line utils that use options and syntax different from what I know (ifconfig, route...) but when I try man, there is no manpage. Should I look somewhere else for the documentation? Or is it just not there yet?
janking has quit [Quit: Vision[]: i've been blurred!]
<qth[m]>
<OscarL> "qth: /boot/home/config/settings..." <- What is the difference between /boot and /haiku?
<OscarL>
/boot always points to the volume you booted from, regardless of its label.
<qth[m]>
<OscarL> "not sure if default install..." <- I have one!
<qth[m]>
OscarL: Oh ok. And /haiku is because that's my partition name?
<OscarL>
right.
<OscarL>
regarding docs... we're a bit of a mess, specially for ported software.
<OscarL>
for git, for example, somehow we default to .html pages when you do "git foobar --help" :-/
<qth[m]>
Yeah I meant more about base system utils
freddietilley has joined #haiku
janking has joined #haiku
<OscarL>
for the ones that are on-tree... if they are not documented on the user-guide... I'm afraid it will be a case of "use the source, Luke" :-/
<qth[m]>
I have been learning freebsd for the last few months and learned to rely heavily on man 😄
<qth[m]>
Right. I have been learning to do that a bit too. :)
<qth[m]>
Thanks for your answers.
<OscarL>
would be nice to have proper man pages for stuff on base system. Maybe one day when we have more users / man-power... one can hope :-)
<qth[m]>
Alright now my setup is reboot-proof. Amazing.
cassisian has quit [Ping timeout: 480 seconds]
<OscarL>
coreutils includes the man pages, at least. qth[m], if you notice some glaring doc omisions, feel free to open a ticket about it...
<OscarL>
either on HaikuPorts (if missing docs are for a 3rd-party command), or over dev.haiku-os.org if from the base system
<OscarL>
if the latter, might not get a proper man page, but maybe we can end up with a page listing differences from other systems, at least.
arti has quit [Remote host closed the connection]
arti has joined #haiku
<OscarL>
concouse shows hrev58933 as "done", but "pkgman update" still shows hrev58930, and it fails to update to it :-/
<OscarL>
*concourse
<OscarL>
guess I'll try updating at another time.
<Begasus[m]>
re
mmu_man has joined #haiku
<Begasus[m]>
bugger, latest Amarok (Qt6) still crashing when playing a song :(
* OscarL
doesn't likes that ShowImage zooms in/out on plain mouse wheel, specially when scrollbars are active.
<Begasus[m]>
I do like that
<OscarL>
zoom should be <some modifier>+mouse wheel instead, IMO.
<Begasus[m]>
could live with that too :)
<OscarL>
I think that if a scrollbar is visible, and "active", wheel should work like on every other app and just scroll. Currently only does that if you move the wheel while having the cursor over the actuall scrollbar :-D
* OscarL
closes ShowImage, before he attempts (*again*) clicking on the close button of a window... from a screenshot :-D
<OscarL>
time to update some drivers on my Win10 install (after it survived the motherboard transplant). Got some shinny VMware Workstation 17 installer to try too.
<OscarL>
(or I might just decide to go to sleep... almost 4:30 down here, heh). See you around Begasus[m], have a great day!
OscarL has quit [Quit: Gone with the wind]
janking has quit [Quit: Vision[]: i've been blurred!]
cassisian has joined #haiku
Aedil has joined #haiku
janking has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
janking has quit [Quit: Vision[]: i've been blurred!]
janking has joined #haiku
janking has quit [Quit: Vision[]: i've been blurred!]
janking has joined #haiku
Nasina has quit [Ping timeout: 480 seconds]
neoncortex has quit []
Nasina has joined #haiku
janking has quit [Quit: Vision[]: i've been blurred!]
HaikuUser has joined #haiku
HaikuUser has quit []
janking has joined #haiku
cassisian has joined #haiku
dodo75 has joined #haiku
dodo75 has quit []
Aedil has quit [Ping timeout: 480 seconds]
dodo75 has joined #haiku
janking has quit [Quit: Vision[]: i've been blurred!]
cassisian has quit [Ping timeout: 480 seconds]
tetrislife has joined #haiku
janking has joined #haiku
talos4 has joined #haiku
talos has quit [Ping timeout: 480 seconds]
talos4 is now known as talos
janking has quit [Quit: Vision[]: i've been blurred!]
janking has joined #haiku
janking has quit []
janking has joined #haiku
cassisian has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
grexe has joined #haiku
<grexe>
call out to the Haiku devs @waddlesplash @mmu_man here: from yesterday to the latest nightly, WLAN breaks here, cannot connect anymore:
<grexe>
getting this error in the syslog: KERN: openbsd wlan_control: 9235, 78 (not supported)
<grexe>
iaxwifi200 driver
<Habbie>
got hrevs?
<grexe>
sure, breaks in hrev58933, works in hrev85926
<grexe>
58926 - sorrt
<grexe>
in other news, I got reverse relations and labelling (classification) working in SEN :=)
<Begasus[m]>
progress! :)
cassisian has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
neoncortex has joined #haiku
tetrislife has left #haiku [Error from remote client]
Nasina has quit [Read error: Connection reset by peer]
Nasina has joined #haiku
cassisian has joined #haiku
cassisian has quit [Remote host closed the connection]
cassisian has joined #haiku
Anarchos has joined #haiku
<Anarchos>
which version of firefox should I install ? pkgman shows me floorp_bin, iceweasel_in, librewolf_bin and waterfox_bin ! 4 prots of firefox are too much.
<Anarchos>
s/prots/ports/
dqk_ has joined #haiku
dqk has quit [Ping timeout: 480 seconds]
Nasina has quit [Read error: Connection reset by peer]
<Anarchos>
hello
<phschafft>
morning.
<Anarchos>
phschafft it is lunch time here in france :)
<Anarchos>
phschafft but morning to you !
<qth[m]>
Bon appétit
<Anarchos>
qth[m] thanks !
Nasina has joined #haiku
freddietilley has quit [Ping timeout: 480 seconds]
freddietilley has joined #haiku
<phschafft>
lunch... hm..
<phschafft>
it's nap time!
janking has quit [Quit: Vision[]: i've been blurred!]
janking has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
Nasina has joined #haiku
<Begasus[m]>
Anarchos: iceweasel works well enough for me*
<Anarchos>
Begasus[m] ok i used falkon but i find it a bit slow to launch
<Begasus[m]>
got some threading issue with Falkon after the move to beta5, so hardly use it
<Anarchos>
Begasus[m] oh i see iceweasel uses 'libcroco' . Must be made by french this very lib :)
<Begasus[m]>
major ones atm* WebPositive and Web
<Begasus[m]>
lol
<Begasus[m]>
who did libdumb? or libass? :P
Nasina has quit [Read error: Connection reset by peer]
<Anarchos>
Begasus[m] iceweasel works smoothly, but cursor is blinking on a black square :(
HaikuUser has joined #haiku
<Begasus[m]>
yeah, that's known Anarchos , same with the other forks
<Begasus[m]>
didn't say it was perfect ;)
HaikuUser has quit []
<Anarchos>
Begasus[m] and @ (french keyboard) and numpads don't work
vdamewood has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
arjen has quit [Ping timeout: 480 seconds]
<Begasus[m]>
I am proud to announce the first CMake 4.1 release candidate. (just got it in over mail) :)
<Begasus[m]>
if no one gets in there I'll dive into it tomorrow :)
<Begasus[m]>
cu peeps!
tetrislife has joined #haiku
<Habbie>
remind me, can i see the offered version with pkgman?
Anarchos has quit [Ping timeout: 480 seconds]
mmu_man has quit [Ping timeout: 480 seconds]
mmu_man has joined #haiku
AndrevS has joined #haiku
tktech28 has joined #haiku
nephele has joined #haiku
<nephele>
kallisti5[m]: there is no nephele_mobile here, so you are talking to an echo... in any case, i've setup my mail server with my own CA, with own Certs, and that is hooked into DANE. technically it is "only" expired, but that should not matter for DANE at all
<nephele>
since using a CA with own trust chain and DANE is a bit uncommon i can see where that may be confusing thouhg. Although, my server *also* accepts mail on port 25... because apple did not support sending encrypted mail, at all... for the webkit bugtracker
<nephele>
Anyhow... will be fixed once i have my new mail server up. Will take a bit of time since i have to learn how opensmtpd works first :D
<nephele>
but so far, much less wierd than postfix!
dodo75 has joined #haiku
<nephele>
luckily, Haiku mail doesn't validate TLS certs at all! It only cares that it *isn't* self signed, so we have the worst of both worlds security there ;)
<Habbie>
lol
<nephele>
I want DNSSEC validation, and DKIM validation on emails
tetrislife has left #haiku [Error from remote client]
<nephele>
when did dovecot get a "dovecot pro" :g
<Habbie>
a quick search of my email archives reveals a Dovecot Pro release email from 2017
<nephele>
i guess i just missed that this is a thing then
<waddlesplash>
nephele: for your focus issues, does your view have ACCEPTS_FOCUS set or whatever that flag is?
<nephele>
yes it does waddlesplash. I can get keyboard events and mouse events for example (can also see them in javascript)
<waddlesplash>
ok
<waddlesplash>
what
<waddlesplash>
's the issue then?
<nephele>
I suspect the issue has to do with the focus, in that webkit does not know wether the view is in focus. You can click on UI controls like links fine, but you can't focus a text field to write int
<nephele>
o
<waddlesplash>
ah
<nephele>
it could also be a problem with something else however
<waddlesplash>
you probably need to tell WebKit when focus changes yes
<waddlesplash>
is that not being done?
<nephele>
There was no explicit code before for this, i tried adding some, but had no effect
<waddlesplash>
the BView hook method is MakeFocus()
<waddlesplash>
that should be called whenever a BView is focused
<nephele>
i don't think this is a hook method?
<waddlesplash>
yes it is
<waddlesplash>
well, it's a hook *and* a perform method
<nephele>
It's only documented under input-related and not hook methods
<nephele>
anyway, that part isn't the problem in that case
<Habbie>
do I see correctly that Genio's LSP support is only for the few languages mentioned in LSPServersManager.cpp?
<nephele>
The problem would be with whatever Function webkit wants called to indicate the focus has changed, that is the part i haven't figure out yet
<nephele>
the view itself does have focus
<nephele>
ugh, do i now use one of those ACME thingies? or make myself a CA again? hmm....
<nephele>
only one provider in the EU
<Habbie>
ACME tends to be the path of least resistance
<nephele>
I like the part of using the critical extension to make my CA only valid for exactly my domain
<nephele>
and that works fine.... except on android
<nephele>
The main reason i did this, in any case, was to be able to receive mail on iOS and Haiku, figuring that if i have a CA that is fine. and the anchor is in DANE anyway so remote servers don't have to care
<nephele>
iOS does not do DANE, and neither does Haiku
<nephele>
unfortunately
<Habbie>
almost nothing does DANE except mail servers with eachother
<Habbie>
(and XMPP servers)
<nephele>
I know. But we can improve this
<nephele>
atleast adding DNSSEC validation for Haiku (with a a unsigned = meh; signed but wrong = connection failed) type of deal
<Habbie>
i know -at least- 3 rabid nerds who would switch OSes to Haiku if we were the one OS to do DANE in our TLS stack for everything
<nephele>
and later DKIM and DANE
<Habbie>
hmm what would DKIM do in Haiku?
<nephele>
you can show in the mail application if an email was actually from the server it claims to come from
<Habbie>
oh in the mail app, gotcha
<Habbie>
i'm not sure that -needs- to wait for DNSSEC support in our resolver lib
<nephele>
yes it does
<Habbie>
DNSSEC support adds value to DKIM, but their is no dependency ordering
<Habbie>
most DKIM in the world today doesn't involve DNSSEC
<Habbie>
s/their/there/
<nephele>
DKIM without DNSSEC is worthless, because you can't get the public key for it without it
<Habbie>
uh no that's wrong
<Habbie>
literally, most DKIM in the world today doesn't involve DNSSEC
<waddlesplash>
our resolver library is mostly just NetBSD's
<Habbie>
right
<Habbie>
so we likely have exposure to the relevant DNS headers bit from a resolver
<Habbie>
which can be enough if you trust it
<nephele>
Habbie: let me put it another way, DKIM without DMARC is worthless, and thus without DNSSEC
<Habbie>
nephele, i still don't follow and i sadly know most of these technologies very deeply
<Habbie>
DKIM, DMARC, SPF are just DNS data
<Habbie>
DNSSEC validates DNS data
<Habbie>
but most DNS data in the world is not DNSSEC protected
<Habbie>
and yet DKIM and DMARC work
<waddlesplash>
nephele: you do realize that Habbie literally works for a company whose primary business is DNS stuff lol
<Habbie>
(you will not hear me say out loud that SPF works ;) )
<waddlesplash>
I think he knows what he is talking about :P
<Habbie>
waddlesplash, i am always ready to go into a conversation in the assumption that "one of us is going to learn something, and i hope it's me" :)
<nephele>
waddlesplash: sorry? what? arguments from authority are worthless. Let me just talk to habbie normally?
<Habbie>
indeed i do not like to argument from authority
<nephele>
Habbie: if you have a DMARC policy my mail server would definetely fail DKIM validation without DNSSEC present
<waddlesplash>
Habbie: fair enough, but I would not be willing to bet money on that :P
<waddlesplash>
nephele: they're not worthless, they're just weaker than all other forms of argument
<Habbie>
although sometimes some context about the person can help others make assumptions about how likely they are to be full of shit
<Habbie>
waddlesplash, this seems like a poor time to bet, yes
<Habbie>
nephele, I must be missing something. do you happen to have docs you can point at that explain the connection to DNSSEC?
<nephele>
Habbie: in regards to DMARC?
<Habbie>
for example :)
<Habbie>
some opensmtpd policy thing perhaps?
<nephele>
Not at hand no. but most of my testing (in that regard) when i was setting up my mail server was from internet.nl. My mail server was postfix and i had it set up to verify incoming DMARC policies IIRC
<waddlesplash>
Habbie: anyway re. DANE. does CURL not support it?
<waddlesplash>
it looks like OpenSSL has functions to enable it
<nephele>
and the only way I know off to validate DKIM using DMARC is to validate the DKIM entry via DNSSEC
<Habbie>
waddlesplash, i don't think so, let me check
<waddlesplash>
Habbie: anyway you can tell those "rabid nerds" that I don't see any reason we can't accept patches to implement DANE in the built-in BSecureSocket (based on OpenSSL, and what email and pkgman use for TLS)
<Habbie>
waddlesplash, ack :)
<nephele>
Wierdly enough it seems the DMARC RFC only strongly suggests that you should deploy DNSSEC, but does not force you too... *ugh*
<nephele>
Habbie: you should not have told me, that just makes me sad
<Habbie>
sorry :/
<nephele>
DANE is nice, but i still think we should validate DNSSEC beforehand
<nephele>
to atleast have the possibility to reject spoofed entries using DANE
<Habbie>
oh yes
<Habbie>
and DANE really expects you to have validated DNSSEC first
<Habbie>
that all said, the *development efforts* are almost independent
<nephele>
yes, true.
<waddlesplash>
Habbie: also re DNS, what I would really like to do someday, is to move the basic DNS resolver into net_server, and then gethostbyname() and other such basic DNS routines then contact net_server, which can do caching, LAN lookups without needing extra plugins, etc.
<Habbie>
you can put DANE in BSecureSocket while the resolver knows nothing about DNSSEC
<Habbie>
it's just a bit of a lie at that point
<Habbie>
waddlesplash, yeah, that's how that appears to be going in most OSes
<nephele>
I think DANE would be great because then we can ignore all the "but i'm just self-signing my cert" kind of email deployments... tell them to just add one DNS entry and you'd be set :)
<Habbie>
waddlesplash, and for bonus points, resolv.conf also points at net_server
<waddlesplash>
it would be really neat if libnetwork.so wouldn't contain a resolver at all
<waddlesplash>
but that may prove too difficult
<Habbie>
you mean, wouldn't know what a DNS packet looked like at all?
<waddlesplash>
right
<waddlesplash>
Habbie: that would also be nice but having net_server behave and respond like a real DNS resolver is probably too much work
<nephele>
our dns resolver is now from netbsd you said, right?
<Habbie>
on a Linux with, say, systemd-resolved, you can have apps find things in DNS without touching the libc bits that know DNS packets
<nephele>
can we use the FreeBSD version that already knows DNSSEC?
<Habbie>
waddlesplash, ack, one thing at a time
<Habbie>
nephele, how much DNSSEC does it know?
<nephele>
well, it has "drill" in the base system?
<nephele>
not sure how "much" that is unfrortunately
<waddlesplash>
netbsd resolver has some DNSSEC code too
<Habbie>
i could have drill in haikudepot in an hour
<nephele>
Habbie: i think they developer that one on FreeBSD? no?
<Habbie>
drill? no
<Habbie>
drill is from nlnetlabs' ldnsutils
<Habbie>
waddlesplash, i think most of them do now, but just for setting and reading some DNS header bits
<nephele>
it seems freebsd uses that now instead of bind
<nephele>
hmm
<Habbie>
uses what?
<nephele>
ldns utils
erysdren has quit [Quit: Konversation terminated!]
<Habbie>
right, they have drill in base but not dig?
<nephele>
yes
<Habbie>
all of that means little for dnssec support in whatever random processes get via libc
<nephele>
"BIND has been removed from the base system. unbound(8), which is maintained by NLnet Labs, has been imported to support local DNS resolution functionality with DNSSEC."
<nephele>
from freebsd 10 release notes
<nephele>
doesn't that mean that local stuff resolves over that?
<nephele>
or is the libc stuff bypassing that?
<Habbie>
it means that it -could-
<Habbie>
if you start unbound, and point resolv.conf at it
<Habbie>
this is also something you could do on Haiku today
<Habbie>
well, if somebody made an unbound port
<nephele>
so, should we? Or try to use some library for net_server? or DIY it?
<Habbie>
i suspect freebsd has made it easy to do this
<Habbie>
nephele, we should, definitely, make it easy to install something like unbound and then also make it easy to use
dodo75 has quit [Quit: Vision[]: i've been blurred!]
<nephele>
I think that if we add this to haiku we should do it in a way that is in the base system and not something you can just turn off
<Habbie>
this is a lot less work than getting validation into net_server (and having DNS run through net_server)
<Habbie>
nephele, oh, people will definitely need a way to turn off DNSSEC - so many networks where you just can't do it
<Habbie>
or at least, can't do it until you made it through the hotel WiFi login screen
<nephele>
we have captive portal recogniation via dhcp
<nephele>
(as a patch)
<nephele>
turning it off seems... very wrong
<nephele>
that's like, not something users should even know exists
<nephele>
(speaking of a desktop OS)
<Habbie>
there are also office situations where DNSSEC can never work
<Habbie>
you definitely can't get away with forcing it on
<nephele>
in what situation can't it work?
<Habbie>
(i am definitely not saying I'm happy about this)
<Habbie>
well, sometimes the enterprise resolver doesn't do DNSSEC, and also you can't talk to outside name servers
<nephele>
isn't the only situation where it doesn't work a hostile DNS server returning wrong info?
<Habbie>
and also many enterprises have internal domains without DNSSEC that may not have been neatly marked insecure in public
<nephele>
I'm not thinking of the situation with unsigned zones
<Habbie>
indeed, there is that -too-
<nephele>
How much of that is relevant to home use?
<nephele>
Can we atleast detect this and tell people that their network is screwed?
<nephele>
(and if not just turn it on silently)
<Habbie>
entire projects have grown and died around this problem
<Habbie>
systemd and dnscrypt-proxy also struggle with it
<Habbie>
all I'm trying to say is that people with more time than you and me have failed to solve this
<Habbie>
systemd-resolved 'solves' things by being very eager to turn off DNSSEC for a (long) while if things fail
<Habbie>
which makes it functional but also opens it up to, as you say, "hostile DNS server returning wrong info"
<Habbie>
but in reality, most servers that "break" or "do not support" DNSSEC are not hostile. just useless.
<nephele>
Well, this stuff is ment to guard against hostile action, not incompetence :)
<Habbie>
yes. the theory is brilliant :D
<nephele>
would be nice if we can atleast have a "this home network has supported it, and it worked fine, now it breaks, this is really SUS"
<nephele>
must have switched routers to telecom, hehe
diver1 has joined #haiku
<Habbie>
DT?
<nephele>
anyway, regardless of all that. It would be great if we have OS level support for all this
<Habbie>
yes
<waddlesplash>
yes, it would
<nephele>
I would love to properly validate DKIM headers in my mail client
<Habbie>
it would be great if people could turn on DNSSEC validation
<Habbie>
it would also be cool to validate DKIM in the mail client
<nephele>
especially for all applications
<Habbie>
again, almost entirely independent projects
<Habbie>
task list for DKIM: (1) make sure we have enough DNS in libnetwork(?) to get DKIM keys (2) do the work in the mail client
<nephele>
haha
<nephele>
first task is write a mail client :D
<Habbie>
uh, ah
<Habbie>
this is a talk about lazy evaluation
<Habbie>
any questions?
<Habbie>
anyway :D
<Habbie>
for DNSSEC, i think a bit of infra to go "there is a validator on 127.0.0.1, please use it" would be the path of least resistance and work
<Habbie>
plus, of course, porting a validator or two
<nephele>
we need the "imap-Fs" like pulkomandy outlines, and then a proper UI for this that can display emails and info, that has a better UI than this hacked together tracker querry
<waddlesplash>
just start by displaying the info in the Mail app
<Habbie>
imap-fs as the main interface to a mail server? that's a cute approach
<waddlesplash>
and worry about Tracker later
<Habbie>
wait is there a Mail app or not
<Habbie>
oh there is
<waddlesplash>
it only displays email messages
<waddlesplash>
it doesn't navigate through email dirs
<waddlesplash>
that's all Tracker
<nephele>
The only thing we have is one to display single mails and send single mails
<nephele>
and a preference for it
<Habbie>
"it only displays email messages" is a delightful sentence. Does it even talk IMAP?
<nephele>
No
<waddlesplash>
mail_daemon does, yes
<nephele>
that's the task of mail_daemon
<Habbie>
ohh
<waddlesplash>
but that is a separate program
<waddlesplash>
it's all part of the base system though
<Habbie>
sorry i was thinking in general functionality, not where it lives
<Habbie>
ok
<nephele>
the imap-fs would be a better userspace interface to mail_daemon than the current "dumping files" aproach is
<nephele>
(not that it can't keep a cache mind you)
<Habbie>
(was just about to say that!)
diver has quit [Ping timeout: 480 seconds]
<zdykstra>
I'm probably in the minority, but I like the way the mail system on Haiku works - though imap in particular could use a lot of love.
<Habbie>
ok there were 3 tasks. DKIM - either in mail_daemon or Mail (we don't say .app, right). DNSSEC - as above. DANE - big task in BSecureSocket
<Habbie>
it sounds like mail_daemon is not unlike a tool such as imapsync
<Habbie>
'here is an imap server' 'here is a dir' 'go'
<zdykstra>
it also handles sending mail, but yup - that's exactly what it is
<Habbie>
ack :)
<Habbie>
some people handle sending mail via IMAP too!
<Habbie>
drop in Outbox, see it go
<zdykstra>
it's pretty fragile and gets into a wedged state where it just silently stops getting new mail messages
<Habbie>
anyway
<Habbie>
if anybody feels like writing a DYI dnssec validator
diver1 has quit [Read error: No route to host]
<Habbie>
first, don't. second, don't. third, i'll help you test it.
<Habbie>
i've written two in my life and neither were very good but the process was enlightening
diver has joined #haiku
<Habbie>
Begasus[m], port request! ldnsutils :)
scantysnax has joined #haiku
<scantysnax>
good morning, good afternoon, good evening, and good night :-)
tetrislife has joined #haiku
<nephele>
zdykstra: regarding what? the technology or the UI?
dqk has joined #haiku
<zdykstra>
what are you asking?
dqk_ has quit [Ping timeout: 480 seconds]
<nephele>
Well, for your comment about liking the way the mail works now :) i.e what parts do you like? The suggested stuff here is not that much of a big departure i think
<nephele>
the only major new thing would be a proper imap viewer that *isn't* a tracker (raw) querry view
janking has joined #haiku
<nephele>
waddlesplash: do we *have* to follow freedesktop for the getSystemIcon api? The first three icons i wanted to suplant from WebPositive don't have good equivalents in the spec
<nephele>
Habbie: maybe you can help me writing a guide instead :P
<nephele>
on how to set up a mail server
<nephele>
oh, by the way.
<nephele>
those mail servers keep sending dmarc reports in like tar archives with json in them or something
<nephele>
it would be super cool to have a mail programm that aggregates that and spits it out (as an email) in human readable form :)
janking has quit [Quit: Vision[]: i've been blurred!]
<Habbie>
i believe there's tools for dmarc
<Habbie>
most of them are not desktop apps i guess
<Habbie>
as for 'writing a guide on how to set up a mail server', no, that's not for me
<waddlesplash>
nephele: what are the icons?
<nephele>
the forward, back, and stop icons
<waddlesplash>
nephele: and I would be surprised if they don't, basically all major icons have FreeDesktop canonical names
<waddlesplash>
yes, those have FreeDesktop versions?
<nephele>
freedesktop spec only had one for stop
<nephele>
for back and forward the only similar one was for listviews...
<nephele>
for next and previous item in a list
<waddlesplash>
those are probably for web browser navigation too
<waddlesplash>
I would check an actual usage of this spec, e.g. KDE or GNOME
<waddlesplash>
I bet Falkon and Firefox both fetch "go-next" for navigation
<nephele>
Let's not do that...this wierd overloading of semantic actions is why so many themes suck
<waddlesplash>
I think the descriptions in the spec are probably wrong in this case
<nephele>
The spec is the spec
<waddlesplash>
I don't know that I have ever seen go-next be used for "list actions"
<nephele>
either we follow it or we don't *shrug*
<waddlesplash>
nephele: "custom has the force of law"
jmairboeck has joined #haiku
<nephele>
not sure what that means
<waddlesplash>
if in fact everyone who supposedly uses the spec actually uses "go-next" for web browser navigation, then functionally, that is what it is
<waddlesplash>
and the spec is a dead letter
<nephele>
Well, in that case, i will simply name that after what they do instead
<nephele>
i.e what their name is now
<waddlesplash>
I fail to see how that makes any sense
<waddlesplash>
if there is a custom for what is to be done here, let's follow that?
<waddlesplash>
but I also fail to see why this is a question right this second
<nephele>
I'm working on webpositive2, webpositive1 has these in it's ressources. But i can add them to getSystemIcon
<waddlesplash>
let's not try and change everything at once
<nephele>
?
<waddlesplash>
if they're in the resources, leave them there for now
<nephele>
I fail to see what that would help
<waddlesplash>
or, remove them from the resources, totally separate from webpositive2 changes
<waddlesplash>
where the icons are has nothing to do with webkit2
<nephele>
No, but webpositive2 is effectively a new browser
<nephele>
so i'm just assembling that
<waddlesplash>
I doubt that it will really be so drastically different
tetrislife has left #haiku [Error from remote client]
<waddlesplash>
during the QtWebKit 1 to 2 transition period there was mostly the same browser shells
<waddlesplash>
our API changes may be bigger but I doubt they will demand a total rewrite of the browser frontend.
<nephele>
UI wise? maybe not. But that isn't the point. If i move part after part i can test every thing while working on the webkit2 changes in tandem
<nephele>
rather than putting it into webkit1 and then having months of regression works where we only have a half working browser
<waddlesplash>
but the thing is that this has nothing to do with webkit2 whatsoever
<nephele>
or having to deal with countless ifdefs
<waddlesplash>
move the icons around in master if you want
<waddlesplash>
but let's keep the big merge commit as small as possible?
<nephele>
what big merge commit?
<nephele>
i'm not touching the code of webpositive(1)
<waddlesplash>
from webpositive 1 to 2?
<waddlesplash>
okay so it'll get replaced, either way?
<nephele>
I'll just develop it side-by-side in another folder
<waddlesplash>
I don't think that makes sense either...?
<nephele>
... okay?
<waddlesplash>
we should have one commit switching WebPositive from WebKit1 to 2
<nephele>
No
<waddlesplash>
if this is a merge commit with smaller commits behind it, fine
<nephele>
I really don't want that
<waddlesplash>
but it doesn't make sense to have two versions at once
<nephele>
Yes it does make sense
<waddlesplash>
Why not?
erysdren has joined #haiku
<waddlesplash>
the APIs are really not that different from Web+'s standpoint
<nephele>
webkit2 is in no state to power webpositive yet. and it will take a lot of work on both sides
<waddlesplash>
OK so let's get it to that state
<nephele>
if i "just" switch now we have a broken webrowser
<waddlesplash>
And I doubt it will really take that much work on "both" sides
<nephele>
And same if we "just" switch is sometime else
<waddlesplash>
Web+ is *not* a small amount of code
<nephele>
I don't care
<waddlesplash>
you are going to rewrite 10,000+ lines of code?
<nephele>
And I don't see why you are opposed to this?
<waddlesplash>
(I just used cloc, it's that much)
<nephele>
I'm not going to rewrite all of it, of course. But I will read all of it. and replace the code that makes sense to do so
<waddlesplash>
nephele: because WebKit2 has been "in the future" for years, and what you are proposing will just push it out further for no good reason
<waddlesplash>
Let's get Web+ switched to WebKit2 as soon as it works "well enough" and doesn't crash all the time. This is what the nightlies are for
<waddlesplash>
We can polish it as we go, rather than waiting around
<nephele>
WTF? just let me work on the damn thing?
<nephele>
I don't see what your problem is with wanting to control other peoples development style
<waddlesplash>
you can do whatever you'd like, but if you open a change copying the entirety of the Web+ code into a directory called "webpositive2", I am likely to downvote it
<nephele>
I'm not going to do that. I'm going to port code of webpositive1 to webpositive2 in incremental commits, and do the required work for webkit2 in the same time
<nephele>
so at first you have a chrome, then the next commit a working tab bar, the next one the cookie bar working etc
<nephele>
with all of these tested and integrated
<waddlesplash>
I think that will likely expand the development process unecessarily and delay webkit2 further
<waddlesplash>
but I can't stop you from doing that, so
<nephele>
not letting me work on it will delay it even more *shrug* :)
<nephele>
In any case, i don't expect webkit2 to be completely ready for beta6. and i'd rather not have one half broken browser
Anarchos has joined #haiku
<nephele>
but that seems inevitable if we just switch it like it is now
<nephele>
there are too many things that work differently for webkit2, and in addition to that we don't even have the api to integrate this into applications ready... so that will probably also have to be done while working on webpositive. to figure out what the api even needs to expose and what not
<nephele>
I mean if you have good reasons why you think my aproach is bad on a technical level then i'm all ears. But it potentially taking longer than a quick and dirty port is not a good reason in my eyes :(
Anarchos has quit []
<waddlesplash>
if things like Cookies are already broken, then switching to WebKit2 won't break them more
<waddlesplash>
so, if we can replace WebKit1 with WebKit2 and not break more things, then let's do that
<waddlesplash>
rather than fixing absolutely everything before switching
<waddlesplash>
obviously more work is needed before we can do that, but not quite so much
Anarchos has joined #haiku
<nephele>
I think reading over all code and integrating it (with additional code review) will give us a much better quality in the long run
<nephele>
and better understanding of why some stuff works or not
<nephele>
rather than the previous cases of something is broken and nobody knowing for months why, because nobody knows how the thing actually works ;)
<nephele>
So i'm not fixing everything before switching. But for your example of cookies, i would suplant the implementation of it in one commit, and then we can fix it, in addition to what webkit2 needs, and the api that is needed for that
<nephele>
and also i'd really like to work on that in parts to prevent beeing burnt out by it, or people in the forum constantly complaining that webkit1 was better because of $issue... :)
<waddlesplash>
people are going to complain no matter what
<waddlesplash>
that's basically inevitable
<waddlesplash>
so I wouldn't worry too much about that particularly
<nephele>
if it's the "new" browser i can tell people to just use the old one instead :)
<waddlesplash>
and as I said, I would vote against having two different "WebPositives" around
<waddlesplash>
If people want the old browser, they can stick with a beta build
HaikuUser has joined #haiku
HaikuUser has quit []
<nephele>
so, what? delay this work after beta6? that makes even less sense
<waddlesplash>
delay what work?
<waddlesplash>
this should proceed as development normally does
<waddlesplash>
namely, at whatever pace it happens to go, and if it's ready for the next release, we merge it
<waddlesplash>
if it's not, it gets merged after
<waddlesplash>
whenever it becomes ready
<nephele>
yes, again, many small tasks to do there. I don't want them in one big commit
HaikuUser has joined #haiku
<waddlesplash>
But many of the tasks you named have no reason to be in a big commit at all
<nephele>
and i don't see what justification you have against this if it's not changing existing code
HaikuUser has quit []
<waddlesplash>
like the cookies one
<waddlesplash>
sure, that should be a small commit. but it also can just wait until after we started using webkit2
<nephele>
Nah
<nephele>
Either we do this with two codebases, or with a whole lot of ifdefs
<waddlesplash>
I obviously can't stop you from doing things some other way, but conversely, you also cannot stop me from doing work either
<nephele>
and i'd rather not work with countless ifdefs
<waddlesplash>
there is another option?
erysdren has quit [Quit: Konversation terminated!]
<waddlesplash>
we just switch WebPositive over in one single commit. I doubt it will be more than +/- 1000 lines
<waddlesplash>
In fact I will be surprised if it is that big
<nephele>
And again. I don't want all work in a single commit
<waddlesplash>
if the total work is +/- 500 lines that's not very much
<nephele>
That is *very* much. because code review will be hours of constant bickering
<waddlesplash>
only if there's things to bicker about :P
jmairboeck has quit [Remote host closed the connection]
<nephele>
You always find something
<waddlesplash>
no?
<waddlesplash>
I usually find things on changes that are API or UX related
<waddlesplash>
the WebPositive change is entirely private API and, for the internals, not UX
<waddlesplash>
so, I don't review those nearly as strictly, because if we get it wrong, we can change it easily without breaking things for others
<nephele>
No, it's almost completely public api
<waddlesplash>
it's not
<nephele>
yes it is
<waddlesplash>
HaikuWebKit is not public api
<nephele>
Yes it is
<waddlesplash>
maybe someday it will be, but it isn't
<nephele>
No, it is, right now
<nephele>
that's the point of webkit2
<waddlesplash>
then declare WebKit2 private/unstable/experimental API
<nephele>
it's not finished right now, but it's definetely public api stuff
<waddlesplash>
it *will be* but *is not yet*
<nephele>
.... so?? Is designing and implementing a public api something that deserves no code review? I highly doubt that
jmairboeck has joined #haiku
<nephele>
yes, most of that code review is in webkit2, but i still consider the work to do here something that should be done at both ends
<waddlesplash>
well that code isn't part of the base system at any rate
<waddlesplash>
it isn't now, maybe it will be in the future but it's just not
<waddlesplash>
it's another package you have to install
<waddlesplash>
so we have a lot more flexibility
<nephele>
And honestly webpostive is too complex for what it's worth that i want to "just" switch it. If you can simplify the code beforehand, for example properly abstracting tabs away and such, we would have a much better base where you could try to "just" change the api
<waddlesplash>
unless the refactor is actually necessary I don't see the point
<waddlesplash>
sure, the code could probably be better
<nephele>
for the tabs the refactor is neccesary
<waddlesplash>
but from my recollection, the code does what it needs to do, I don't recall seeing "unnecessary complexity"
<waddlesplash>
okay, but it sounds like Genio maybe already did that work?
<waddlesplash>
so we don't even have to
<nephele>
for one, no that won't be sufficient, and for another you yesterday argued against putting their implementation in libshared
<waddlesplash>
no
<waddlesplash>
I argued against putting any code in libshared *that nothing was actively using*
<waddlesplash>
if you import the code and use it in the next commit, that's fine
<nephele>
Well, what we'd basically need, I think. Is something based on BCardLayout that can be controlled externally, with tab headers or a list
<nephele>
does Genios refactor fit that?
<waddlesplash>
I don't know, I have not looked at it
<waddlesplash>
but I used the Web+ tab view to do something like this in Heidi years and years ago
<waddlesplash>
and IIRC it already more or less worked the way I needed it to
<waddlesplash>
so, I don't think this would be as difficult as you seem to think
<nephele>
I "just" think the implementation is shoddy, and wouldn't work well with webkit2 as is
mmu_man has quit [Ping timeout: 480 seconds]
<nephele>
no answer from codeberg yet, still...
<Habbie>
how do i get pthread_setname_np? i see it in headers/compatibility/gnu/pthread.h but i'm unsure this is indirectly included by default
<Habbie>
ah, a bunch of haikuports just patch the call out
<nephele>
waddlesplash: in any case, regardless of which aproach we use. webpositive would still need some big work to become a better browser
<waddlesplash>
Habbie: it is indirectly included. Should be automatically enabled under _DEFAULT_SOURCE
<Habbie>
i saw that guard
<waddlesplash>
which should be enabled so long as you didn't use -std=c99 or something
<Habbie>
meson is not setting the define
<Habbie>
ah
<Habbie>
-std=c++17
<waddlesplash>
use -std=gnu++17 instead
<Habbie>
right
<waddlesplash>
that avoids defining "strict mode"
<waddlesplash>
nephele: yes, I agree of course
<Habbie>
ok that helped
<nephele>
ha! I fixed the flickering
pabs has quit [Remote host closed the connection]
<nephele>
me: changes nothin; webkit: random build failure in unrelated file
pabs has joined #haiku
<Anarchos>
nephele :)
HaikuUser has joined #haiku
HaikuUser has quit []
nephele has quit [Quit: Vision[]: i've been blurred!]
Anarchos has quit [Ping timeout: 480 seconds]
dalme has joined #haiku
Anarchos has joined #haiku
Nasina has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
HaikuUser has joined #haiku
HaikuUser has quit []
Anarchos has quit [Ping timeout: 480 seconds]
<Habbie>
genio nit: it actively complains when the project i open is -not- a git repo
jmairboeck has quit [Read error: Connection reset by peer]
stux- has joined #haiku
<Habbie>
Errno -2147459069, python is confused by something
Nasina has quit [Remote host closed the connection]
AndrevS has quit [Quit: umount /dev/irc]
dalme has quit []
Nasina has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
Nasina has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
nephele has joined #haiku
<nephele>
Haha! I can now focus text fields in webkit
<phschafft>
huray!
AlaskanEmily has joined #haiku
xet7 has joined #haiku
<nephele>
Darn, I can now focus text fields. so the focus issue is resolved, and i know that in javascript key events arrive properly.... but i still can't write into textfields
<nephele>
what's missing :?
ChaiTRex has quit [Remote host closed the connection]
ChaiTRex has joined #haiku
erysdren has joined #haiku
Nasina has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
Nasina has joined #haiku
Nasina has quit [Read error: Connection reset by peer]