<phschafft>
hope the work on the XMPP side is kept going.
<phschafft>
discussed a bit integration with other parts of the system there.
<Begasus>
Don't think progress on Renga will stop any time soon
<phschafft>
my very basic point is: of e-mail is part of the system, then so should other communication protocols as well.
<phschafft>
e.g. one of the topics we discussed was that we have person files for contacts used by e.g. e-mail. but the buddy list is stored in a different way.
<Begasus>
in the "people" files?
<phschafft>
that doesn't require much updates on the base system, still would get a good deal into the right direction.
<phschafft>
Begasus: hm?
<Begasus>
we already got those, maybe add the additional information in there?
<phschafft>
yes, that is what I was suggesting.
<Begasus>
on par then :)
<phschafft>
to migrate the buddy data from XMPP to live in the person files as well.
<phschafft>
we also discussed the details of some of the data structures, both logically and on-disk format.
Nephele has joined #haiku
* phschafft
waves to nephele.
<Begasus>
Hi nephele
<phschafft>
was just talking about you! ;)
<Nephele>
oh?
<phschafft>
our meeting yesterday.
<Nephele>
phschafft: fwiw email is an originally third party app for BeOS imported into Haiku,
<Nephele>
and as such rewritten partially over and over to make it more native
<phschafft>
it feels very native. ;)
<Nephele>
for Renga the client to xmpp is written in gloox. it's competent, but sadly also some gpl variant. so for the base system it is not really great
<Nephele>
s/in/with/s
<Nephele>
phschafft: sure it is native! but more integration would be cool. I think logging is one thing i want to tackle. I don't like the current system log of "it's just a text file"
<Nephele>
especially since that text file contains color control sequences for terminals
* phschafft
blinks.
<Nephele>
mail_daemon logs to stdout or stderr
<phschafft>
storing ANSI control sequences is on the same level as dumping your memory as-is to the disk.
<phschafft>
nephele: I'm also thinking about doing some work on logging in SIRTX and Manuel already started with me doing a bit in Data::URIID. maybe we can get all three a bit in logic sync.
<phschafft>
I mean we don't need three people indpendently solving the same problem differently at the same time. ;)
mmu_man has quit [Ping timeout: 480 seconds]
tqh has joined #haiku
<Nephele>
phschafft: 4 people actually ;)
<phschafft>
oh?
<Nephele>
and the 4th one has pretty much already solved this problem, so i will make my design mostly compatile
* phschafft
wonders if nephele has started to count the voices in someone's head.
<Nephele>
the crash reports are basically "just json" with the log viewer having rendering code if you want, and beeing able to querry for certain log lines and such
<Nephele>
also having a stack of recent filters/searches, you can switch between these states directly etc
<Nephele>
it's made by the author of the iOS xmpp client Monal.
<Nephele>
This is one thing i learned in the coding sprint :D, also mentioned this in my blog post
<Nephele>
(or well, this software is not made by him directly but his student iirc)
<Nephele>
maybe I can ask him if he'd be interested to join a call to share his stuff :)
<Nephele>
if you'd like
<phschafft>
hm.
tuaris has quit [Read error: Connection reset by peer]
<phschafft>
I see.
mmu_man has joined #haiku
<phschafft>
another topic that is on my mind for some days:
<phschafft>
I'm writing this little network based Tic Tac Toe client. was wondering if anyone has interest in creating a version for Haiku.
<Nephele>
of those protocols listed only xmpp seems to still be available :3
mmu_man has quit [Ping timeout: 480 seconds]
mmu_man has joined #haiku
JakeSays1 has joined #haiku
JakeSays has quit [Ping timeout: 480 seconds]
<ilzu[m]>
Yup. And I guess the whole codebase has bitrotten to the core, but the idea phschafft described sounds pretty similar
walkingdisaster has joined #haiku
walkingdisaster has quit []
* phschafft
nods.
<phschafft>
sounds a bit like it.
* phschafft
clearly has his hidden agenda of later on adding support for his database based chat protocol ;)
<Nephele>
phschafft: should i ask tmolitor about a chat about the logging system?
<phschafft>
I'm happy for anyone interested to join our meetings. I'm also happy for anyone joining this channel. ;)
<phschafft>
but I'm not sure (until I looked at what you linked) if we're actually looking to do the same basic thing.
<phschafft>
logging vs. debugging.
<phschafft>
(with a nice, blurry border between them ;)
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
<Nephele>
well, this does both kinda. You can also use live logging over the network with it, to debug the application
<Nephele>
but it also can read logs that the application was writing normally (and then crashed for example)
mayuri has joined #haiku
<phschafft>
that is both the logging area for me.
<phschafft>
the difference is more about what data is part of the log stream. loggging for me is more about stuff the user can work with, and debugging is more about stuff that makes sense to a developer.
<phschafft>
so for example a backtrace is debugging, while a cannot-open-file-X message is logging.
<Nephele>
to me your distinction doesn't make sense. logging is a part of debugging, not seperate to it
erysdren has joined #haiku
<Nephele>
i only need logs to figure out why or how something goes wrong
<phschafft>
it's a very blurry border.
<Nephele>
user visible (fixable) errors should not need people to view the debug "full" log :)
<phschafft>
nephele: logging can also be required to ensure things can be understood later. e.g. for auditing.
<Nephele>
i.e the mail_daemon on haiku has a user visible error log, but it only sais stuff like "this email cant be send, your TO: adress is wrong" and such
<Nephele>
sure, i suppose
<phschafft>
so not all logging needs to be about errors to begin with.
<Nephele>
the mail_daemon also has a second seperate log where it spits out *everything* it is doing
<phschafft>
access log and accounting is also logging.
<phschafft>
maybe you don't want to call it different things but maybe you want to call it having a different focus.
<Nephele>
sure. I think these should be collected by the same server in haiku
dodo75 has joined #haiku
<phschafft>
most likely.
tqh has quit [Remote host closed the connection]
HaikuUser has joined #haiku
HaikuUser has quit []
DKnoto_W has quit [Ping timeout: 480 seconds]
DKnoto_W has joined #haiku
zard has joined #haiku
zard has quit []
<phschafft>
found a memory access error in SIRTX on archs that share the same address space for read only and read-write memory.
<phschafft>
fun.
<erysdren>
like what archs?
flag has joined #haiku
<Habbie>
most archs, i think?
flag_ has quit [Ping timeout: 480 seconds]
<phschafft>
what Habbie said basically.
<phschafft>
I would say at least most von Neumann archs. which is the majority at least by how much they are known.
<Habbie>
i'm not even sure i've ever worked with anything different
<Habbie>
except on paper, in university ;)
<phschafft>
think about your caluclator.
<phschafft>
if you have a classic one, not one of those modern ones that run android, than it's likely harvard arch.
* phschafft
still has on his list to show Habbie what he is working on, in case Habbie is interested.
zard has joined #haiku
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
HaikuUser has joined #haiku
<HaikuUser>
hi
<zard>
Hello HaikuUser :)
<zard>
You can set your name by typing "/nick newname"
HaikuUser is now known as mr
<Begasus>
k, time to checkout new netbeans ...
<mr>
im drinking cola
<Begasus>
Hi zard, thanks for the screenie! :D
<zard>
:D
<mr>
dameopyyyyyy
<zard>
?
<Begasus>
download + 478MiB, not that bad :P
<mr>
usa
leahc has quit [Remote host closed the connection]
toasterking has quit [Remote host closed the connection]
dominicm has quit [Remote host closed the connection]
utzig has quit [Remote host closed the connection]
wolfdog has quit [Remote host closed the connection]
rodolphoeck has quit [Remote host closed the connection]
<Begasus>
Up and running together with latest LTS nodejs, NetBeans IDE22 :) https://0x0.st/XpVu.png
leahc has joined #haiku
<Begasus>
included composer.js installed with latest npm :D
<Begasus>
included javafx?
toasterking has joined #haiku
_-Caleb-_ has left #haiku [#haiku]
dominicm has joined #haiku
rodolphoeck has joined #haiku
_-Caleb-_ has joined #haiku
Begasus_32 has quit [Quit: Vision[]: Gone to the dogs!]
DKnoto_W has quit [Quit: Leaving]
DKnoto has joined #haiku
<Begasus>
[notice] A new release of pip is available: 23.0.1 -> 24.1.2 -> OscarL :)
<OscarL>
just wake up into some IRL complications. Will sort out things best I can for now, and probably give that "pip" a try as a distraction (after some coffee).
HaikuUser has joined #haiku
<Begasus>
ouch, no hurries OscarL, was just pulling your leg (as usual)
* phschafft
offers OscarL a cookie, also happy to provide a hug if needed.
<OscarL>
Thanks Begasus, phschafft. /me waves zdykstra, while he's at it.
<OscarL>
(after a quick look at the logs...)
<OscarL>
I wouldn't considered Mail "3rd Party software imported into Haiku". Big chunks of Haiku servers/kits/drivers/apps started as "for BeOS replacements until OpenBeOS is ready" projects. MDR == "mail daemon replacement", and at least some of its authors were also OpenBeOS maintainers.
dovsienko has joined #haiku
<OscarL>
Begasus: managed to fix that npm error on 32 bits already?
* OscarL
remembers replacing servers one by one on his BeOS install with OpenBeOS ones, as soon as they were available.
MisthaLu has joined #haiku
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
<Nephele>
OscarL: wasn't BeMail available before OpenBeOS?
HaikuUser has joined #haiku
HaikuUser2 has joined #haiku
HaikuUser has quit []
HaikuUser2 has quit [Remote host closed the connection]
<Nephele>
or was this split, that some parts were third party and others in BeOS proper?
<OscarL>
BeOS had its mail client app, and server, sure. that'ts what MDR set to replace.
<OscarL>
(just in case, my info/experience is from using R5 onwards, not sure how things were before that)
<Begasus>
OscarL, npm not available in 32bit
<Begasus>
no nodejs :/
<Nephele>
hmm, mail kit was in beos too
<OscarL>
Begasus: just asking because the error you posted earlier doesn't seems too hard to fix (just needs a cast).
<Begasus>
ah, that waas from the test run on 32bit yes
<Begasus>
with latest LST
<OscarL>
changing line 78 to: "info.GetReturnValue().Set((uint32_t)uv_os_getppid());" <<< might be enough?
<Begasus>
different error then with previous version and with the nodejs16 one
<Begasus>
booting 32bit back up
BrunoSpr has joined #haiku
HaikuUser has quit [Remote host closed the connection]
<Begasus>
build launched, takes a bit though ;)
<Begasus>
error in netbeans trying to install pylsp deps: https://bpa.st/WXKQ :)
BrunoSpr has quit []
<Begasus>
probably our pyzmq is too old for that
BrunoSpr has joined #haiku
<OscarL>
mmm, I did have python-lsp-server working witouth issues last time I've tried with CudaText. But haven't tried to (re)install it lately.
<Begasus>
not a big deal for now, guess I tried to install too much plugins at the start :D
<Begasus>
getting nodejs20 build on 32bit would be a big plus
HaikuUser has joined #haiku
<OscarL>
(for CudaText, I've installed the python lsp server with pip at the time, not wanting to work on packaging it :-D)
HaikuUser has quit []
<Begasus>
how did you install it?
<OscarL>
(booting up Haiku, to see if it is still there)
<Begasus>
tss :)
<OscarL>
pip list shows: "python-lsp-server 1.7.4"
<OscarL>
so still there at least.
tqh has joined #haiku
<Begasus>
ok that seems to work
<Begasus>
nogo in netbeans :/
<Begasus>
biab, will let nodejs20 fail gracefully and try to tackle then, thanks on the pointer OscarL
<OscarL>
np.
nosycat has joined #haiku
BrunoSpr has quit [Quit: Vision[]: Ich wurde eingeweicht!]
mmu_man has joined #haiku
<OscarL>
Begasus: FWIW, "pip install -U python-lsp-server" upgraded me to 1.11.0 with no problems. I don't have pyzmq installed. Maybe that "pyslp" is a different fork?
<OscarL>
*pylsp
<OscarL>
Begasus: re: "error: call of overloaded 'Set(uv_pid_t)' is ambiguous"... uv_pid_t seems to be "typedef pid_t uv_pid_t", and on Haiku: "typedef __haiku_int32 pid_t;", so...
<OscarL>
"info.GetReturnValue().Set((int32_t)uv_os_getppid());" might be actually better.
mmu_man has quit [Ping timeout: 480 seconds]
dodo75 has quit [Quit: Vision[]: i've been blurred!]
dodo75 has joined #haiku
dodo75 has quit [Quit: Vision[]: i've been blurred!]
dodo75 has joined #haiku
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
<Begasus>
now you tell me :P
<OscarL>
coffee was not strong enough earlier, sorry :-)
<Begasus>
let's see how far this goes, thanks!
<Begasus>
biab (doggies)
bbjimmy has joined #haiku
jmairboeck has joined #haiku
<Begasus>
still running (not sure now if it used the first or second sollution) :)
<Begasus>
Hi jmairboeck
<OscarL>
I *think* that at least for testing, the first one should also work (assuming no pid_t values are negatives).
<OscarL>
but... better the second for "formal" .patchset (assuming that's all what is needed to make it build).
<OscarL>
(looking at updaging pip here now)
<Begasus>
+1
<Begasus>
pip list lsp doesn't show any pyslp
<OscarL>
"pip list" shows "python-lsp-server" for me, no "lsp"
<Begasus>
yeah and python-lsp-jsonrpc
<OscarL>
I would need to see what netbeen is trying to "pip install"
<OscarL>
mmm, no more "setup.py" on pip tarball... this will be "fun" :-(
<OscarL>
from your log... problem is not with python-lsp-server, but from some of the other plugins.
<OscarL>
(given that I *do* have python-lsp-server 1.11.0 installed without pyzmq)
<OscarL>
"pylsp-mypy" seems overkill.
<Begasus>
yeah, cmake can't seem to find "python" in the end
<Begasus>
"pip install -U pyzmq" fails
<OscarL>
not sure what requires it.
<Begasus>
not even going to break my head over it :P
<Begasus>
the problem is with these /tmp directories, you can't "fix" it there ;)
<OscarL>
projects assuming *now* python exists as python3 are dumb.
<Begasus>
with you on that one ;)
<OscarL>
Begasus: can't you disable Netbeans trying to install things, and letting the user sort things out? (it installing an unknown set of dependencies as part of its build process is also crazy)
<Begasus>
I think this is because I enabled the python plugin OscarL :)
<OscarL>
Still pretty stupid on their part, IMO.
<Begasus>
well, it's a _bin package, so not much we can do there "source wise"
<Begasus>
not going nuts over it (more then already is) :)
<Begasus>
but now pyzmq is biting me because of this :P
<nekobot>
[haiku/haiku] 372509626a56 - kernel/riscv64/arch_debug: port from x86 version
<OscarL>
latax2html broken for 32bits? "requires "cmd:pnmcrop" of package "latex2html-2024-1" could not be resolved" (from report.txt)
<OscarL>
SECONDARY_ARCHITECTURES="x86_gcc2 x86" <<< on netpbm-10.86.42.recipe sounds weird, and resulting packages do not provide any cmd: on 32 bits...
<OscarL>
that what my last changes, but still missed some.
<waddlesplash>
that should probably be "platbase"
<waddlesplash>
not "base", yes?
<OscarL>
all give same results on Haiku (last time I've tried to figure that out)
<waddlesplash>
ok
<waddlesplash>
well, one of them needs to not be non-packaged
<waddlesplash>
it should probably be the "plat" one
mmu_man has quit [Ping timeout: 480 seconds]
<OscarL>
thus why I even asked nephele to run some sysconfig script for me on OSX..., I have little to compare, so also why I had to setup a linux machine.
<OscarL>
have made some progress on a new .patchset, but I'm slow, dumb, and with terrible focus issues :-D
mmu_man has joined #haiku
FreeFull has quit []
FreeFull has joined #haiku
yann64 has joined #haiku
yann64 has quit []
yann64 has joined #haiku
yann64 has left #haiku [#haiku]
gouchi has quit [Remote host closed the connection]
<OscarL>
great, seems I managed to wipe my WIP .patchset rework stash somehow. Sigh.
MisthaLu has quit [Quit: Leaving]
<OscarL>
at least I still have the data I was collecting while trying to figure out those dirs: https://bpa.st/ZTNA
<OscarL>
yay for me being paranoid about borking git stuff, and havign a .bak file at least :-D
jmairboeck has quit [Quit: Konversation terminated!]
<erysdren>
im playing around in VGA text mode under MS-DOS, creating a BeOS styled GUI: https://0x0.st/XpJ8.png
<OscarL>
waddlesplash: just noticed your "python3.10-3.10.13" (emphasis on .13), tried python3.10-3.10.14 already?
<waddlesplash>
OscarL: no, I didn't. let's see what happens
<waddlesplash>
OscarL: it lists platinclude but platinclude is non-packaged
<waddlesplash>
that should be regular packaged dir
<OscarL>
alright... adding that for next patch :-)
<waddlesplash>
OscarL: ... or should both of them be regular packaged dir?
<waddlesplash>
"python3 -m sysconfig" on Linux doesn't list anything in /usr/local
<OscarL>
still get dizzy with all these different install schemes, to be honest (thus why I was colllecting info on other OSes)
<waddlesplash>
I guess we might as well have include=non-packaged and platinclude=(packaged)_
<waddlesplash>
that makes sense to me
<waddlesplash>
and will fix the problem with GDB :)
<OscarL>
(I usually just "pprint.pprint(sysconfig.get_paths(scheme))", or "pprint.pprint(sysconfig.get_config_vars()))")
<OscarL>
will try to keep that in mind, and will "CC:" you when I open the next Python PR, just in case :-)
<OscarL>
"base vs platbase" is "_PREFIX vs _EXEC_PREFFIX", but there's also "installed_base vs installed_platbase"... with no much documentation on the intended purposes (that I could find at least).
<nekobot>
[haiku/haiku] 65a86bce722a - kernel/fd: use function table to identify file descriptor type
<OscarL>
waddlesplash: default install scheme is "haiku", that goes to non-packaged, scheme "haiku_vendor" should have correct platincludes (on 3.10.14 at least).
<waddlesplash>
that scheme isn't the default, though
<waddlesplash>
so GDB would still need to be patched...
<OscarL>
"export HAIKU_USE_VENDOR_DIRECTORIES=1" should make "haiku_vendor" be the default.
DKnoto has quit [Ping timeout: 480 seconds]
<waddlesplash>
ah
<waddlesplash>
OK then
<OscarL>
for more info... ask the big guys... korli/PulkoMandy, please... I'm just a little mouse :-)
<OscarL>
joking-not-so-joking aside... I'll try cleaning up things further at least, so thanks for helping me understand these things waddlesplash.
<waddlesplash>
if we have that export HAIKU_USE... then using the vendored directories may not be necessary?
<waddlesplash>
I mean, using the vendored directories in the default output
<waddlesplash>
but I guess I'm not sure how this is used
<OscarL>
"vendor" scheme is for python packages installed from .hpkg, then we have "haiku" for non-packaged (but system), then there's "user" (also non-packaged), and home... pretty much broken ATM.
<OscarL>
pip can only install under non-packaged, of course, thus why the defaults go there, I guess.
_-Caleb-_ has left #haiku [#haiku]
<OscarL>
("home" being for .hpkg intalled via "pkgman install -H", unless I'm missing something else)
zard has joined #haiku
_-Caleb-_ has joined #haiku
<OscarL>
hard to match "posix_home" with "haiku_home" intended usages for me (given that posix_user exists).
DKnoto has joined #haiku
* OscarL
takes a look at erysdren's screenie. NOICE! :-)
<erysdren>
danke :3
* OscarL
has been tripping over _INSTALL_SCHEMES related bugs since he first tried to "pip install pyserial".
juanjo has left #haiku [Disconnected: Hibernating too long]
<OscarL>
I just wanted to do some quick tests for my serial_mouse driver couple of years ago... now doing .patchset cleanups, lol :-)
HaikuUser has joined #haiku
<HaikuUser>
hi
HaikuUser is now known as Sparkle
<OscarL>
erysdren: I'd totally use a text-mode-only Haiku :-)
<OscarL>
phschafft: do not distract the man until R1 at least, please.
<phschafft>
OscarL: I always try to stay out of the way of people actually working, and in the way of people wasting time.
<OscarL>
excellent!
<phschafft>
(and I really hate those peopl that think just because they are a customer staff in a supermarket needs to move around them...)
* coolcoder613
waves
* phschafft
waves back to coolcoder613.
<phschafft>
this struct contains an union containing a struct containing an union containing a struct.
<phschafft>
and for some reason it still looks fine, relatively simple to read, and generally valid.
<phschafft>
but I'm sure at some point someone will complain and tell me that I should just have declared a range of bytes and then go with no types at all or convert everything into a large struct with must feels being exlusive to each other.
<erysdren>
BeOS style windowing stuff in VGA text mode.
<coolcoder613>
Nice!
<coolcoder613>
Now you need to multitask it, to get a time machine... then you can sell it
<erysdren>
so true lmao
<phschafft>
;)
<phschafft>
I mean there were products like that.
<OscarL>
build systems guessing a package version from the presence (or not) of a .git folder, when you got the sources from a release tarball... sigh.
<phschafft>
what fun.
* coolcoder613
decides to tell everyone about what lengths he is going to to install Fiwix