WildKeccak has quit [Read error: Connection reset by peer]
HaikuUser has joined #haiku
HaikuUser is now known as WildKeccak
WildKeccak has quit [Quit: Vision[]: i've been blurred!]
Udi has quit [Quit: Vision[]: i've been blurred!]
Diver has quit [Quit: Leaving.]
bjorkint0sh has joined #haiku
bjorkintosh has quit [Ping timeout: 480 seconds]
Babaj has quit [Quit: Leaving]
bjorkint0sh has quit [Ping timeout: 480 seconds]
Maturi0n has joined #haiku
Maturi0n_ has quit [Ping timeout: 480 seconds]
AlaskanEmily has joined #haiku
<nebulabc>
I've been using Haiku on and off over the last year, and really enjoy it on some of my lower end hardware. That said, does anyone know how to make the "scroll button"/middle click on thinkpads work? This is a killer feature for me, and I looked on the fourms and couldn't find anything conclusive.
<nebulabc>
some devices like thinkpads in the x200 line don't even have a trackpad. so that means that scroll bar is the only option. Luckily most of mine have it where you can scroll on the side of the trackpad, but this isn't a perfect solution
OscarL has joined #haiku
<OscarL>
nebulabc: How's that "scroll/middle button" handled in other Linux / *BSD ? If it needs a custom driver, Haiku might just not have it yet. If it is handled by PS/2 (or USB HID) on other OSes... maybe Haiku's drivers just need to be adjusted for that particular hardware.
<OscarL>
In any case... you're best bet is filing a ticket over https://dev.haiku-os.org (search first for your hardware model, to see if something's already there).
<OscarL>
("handled in other Linux / *BSD" I was about to write "in other OSes", and forgot to remove "other" when decided to be more specific :-D)
<OscarL>
Alternatively, if you know how to program, creating an input filter that lets you simulate an scroll wheel (using other keys), is not really that hard.
floof58 has quit [Ping timeout: 480 seconds]
floof58 has joined #haiku
<nebulabc>
I'm pretty sure I saw someone in a forum post say it was something to do with the PS2 implementation. I've had it work without any special drivers.
<nebulabc>
It
<nebulabc>
It's just so you can scroll with the trackpoint by holding that mouse button
<nebulabc>
right now it just treats it as button 2 in a 3 button mouse
<OscarL>
In that case, I strongly suggest opening a ticket about it. Things "get lost" in IRC/forum, and the devs prefer to have a ticket to track such issues.
<augiedoggie>
should be pretty easy to write a filter for that
<augiedoggie>
i've seen similar-ish ones on github
<Begasus>
forgot to strip the package for iaito I saw :/
<Begasus>
before about 50MiB, after debuginfopackage created about 2MiB :)
<OscarL>
ouch.
DKnoto has joined #haiku
<Begasus>
already had it (for checking out git commits)
<Begasus>
and being stuck with lua stuf :/
<OscarL>
Been doing some little benchmark compares on my Phenom II X4. Linux vs Haiku on both pystone.py, and "syscall-overhead"... using both bare-metal and VBox on Win10.
<Begasus>
and?
<OscarL>
my "syscall-overhead" results for Linux / Haiku on VBox are really crazy. Linux on VBox: 124 ns (138 ns on bare metal, LOL). Haiku on VBox: 3600 ns (158 ns on bare metal).
<OscarL>
yet.. when comparing python's performance, I only see a ~12% difference on favor of Linux.
<OscarL>
Not bad. (same for 3.10, and the faster 3.11)
<Begasus>
quite some big difference on the first check
Begas_VM has joined #haiku
<OscarL>
yeah... VBox+Haiku is nuts for syscall-overhead, yet, the system runs fine.
<Begas_VM>
that's the main part +1 :)
<OscarL>
indeed.
<OscarL>
Really surpriced that our pythons are now only 12% slower that on Linux.
<OscarL>
Fixing those OPT flags really helped, it seems :-D
<Begas_VM>
the right guy to the job! +1
<OscarL>
only took me 8 months to find out there was a problem there! /me is slow.... wonder if I'm also missing some OPT flags :-P
<Begasus>
got some things here that even after a few years haven't been able to complete :)
<OscarL>
I took a look at Lua's packages a few weeks ago (mostly to fix that vis/lua issue on 32 bits)... but got distracted, and haven't done any real progress there :-(
<Begasus>
there iaito announced at the forum :)
<Begasus>
I'm still a bit unsure where lua checks for the modules etc ...
<OscarL>
yeah, and I've already forgot what "path" thing I was about to try to fix for them (should check the logs some day).
<Begasus>
also switched here already to default lua5.4 (to be sure things work as expected), lua seems fine there
<Begasus>
I also see that for instance luacheck still defaults to lua5.1? :P
<Begasus>
or was it luafilesystem ...
<OscarL>
we need some lua dev to help here... me looks around... and points fingers...
<Begas_VM>
it took long enough to have somewhat of an python guy to step up :)
<OscarL>
now I understand why some projects reject adding stuff unless it comes with a maintainer attached :-D
<Begas_VM>
yeah, the guy from iaito asked yesterday "how many projects do you maintain?" ... can't count the numbers :) (and I'm not even a dev)
<OscarL>
"ALL OF THEM!!! Send HALP!!!" -- Begasus, probably.
<OscarL>
and that's why I say we're really lucky to have you :-)
<Begas_VM>
lol, thanks, otherwise korli would have a fulltime job :)
<OscarL>
Heh... robots don't count! :-P
<Begas_VM>
heh
* OscarL
looks around to see if korli has logged in already.
<Begas_VM>
he's around the corner :)
* OscarL
hides behind some python test runs results.
<Begas_VM>
probably already read those :P
<OscarL>
Good, doubt he'll look again there, then! :-P
freakazoid332 has quit [Ping timeout: 480 seconds]
frkazoid333 has quit [Ping timeout: 480 seconds]
freddietilley has quit [Quit: WeeChat 3.8]
freddietilley has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
ScottD10 has joined #haiku
ScottD1 has quit [Read error: Connection reset by peer]
ScottD10 is now known as ScottD1
dryphb has joined #haiku
<Begas_VM>
keeping the same path for luafilesystem's recipe
<Begas_VM>
should I just go for default lua5.4, or start with creating subpackages for the other versions ...
<pairisto[m]>
is there anyone who has had the problem of when they modify a file and then go to rebuild haiku using jam -q -j2 \@nightly-anyboot and then the change isn't reflected when you run the file?
<pairisto[m]>
s//`/, s/\@/@/, s//`/
frkazoid333 has joined #haiku
<pairisto[m]>
the only way that I have found to remedy this is to rm -rf generated but it then takes way to long to rebuild using ./configure --cross-tools-source ../buildtools --build-cross-tools x86_64
trungnt2910[m] has joined #haiku
<trungnt2910[m]>
pairisto[m]: Instead of doing that try touch `**/*` from the root of the source.
<trungnt2910[m]>
Or, if you aren't making changes to the kernel itself, try building only the target you modified instead of the whole image.
<trungnt2910[m]>
> <@pairisto:matrix.org> the only way that I have found to remedy this is to `rm -rf generated` but it then takes way to long to rebuild using `./configure --cross-tools-source ../buildtools --build-cross-tools x86_64`
<trungnt2910[m]>
* Instead of doing that try `touch **/*` from the root of the source.
<trungnt2910[m]>
trungnt2910[m]: Or `rm -rf generated/objects` instead. This will keep the cross-compiler intact.
dryphb has quit [Remote host closed the connection]
<nekobot>
[haiku/haiku] 426dc0c7e44a - kernel/arm: remove unneeded call to arch_thread_set_current_thread
HaikuUser has joined #haiku
HaikuUser has quit []
AlaskanEmily has quit [Remote host closed the connection]
<Begas_VM>
well ... it works :) (but hardly usable for others as the devel packages conflict) :)
<pairisto[m]>
so `rm -rf generated/objects` didn't work, since I am making changes to a driver I couldn't necessarily rebuild it but I tried `jam -q -j2 "<driver>tun_driver"` and after rebuilding it it still didn't refresh, and neither did `touch **/*` :/
dryphb has quit [Remote host closed the connection]
<OscarL>
pairisto[m]: by "didn't refresh" you mean that "tun_driver" (and driver.o) under "generated/objects/haiku/x86_64/release/add-ons/kernel/drivers/misc/" do not change at all?
korli has joined #haiku
<korli>
pairisto[m]: you should be in the generated directory when calling jam
<korli>
then "jam -q <driver>tun_driver" should update files in generated/objects/haiku/x86_64/release/add-ons/kernel/drivers/misc/ like OscarL wrote
<korli>
to update the image at the same time, jam -q update-image @nightly-anyboot "<driver>tun_driver"
mmu_man has quit [Ping timeout: 480 seconds]
smalltalkman has quit []
<PulkoMandy>
Also, just to make sure: rebuilding a driver like this updates it in the generated/ directory, but does not update the generated filesystem image if that's what you're testing with
<Begas_VM>
current luafilesystem installs "lfs" in $libDir/lua5.3, luv installs the shared libraries in $libDir, but the library "luv" in $prefix/lib/lua5.3
<Begas_VM>
luafilesystem has support for both arch's, so can't use $prefix/lib there ...
<OscarL>
is gcc2 suppport really necessary for modern Lua versions? "Lua 5.0 was released on 11 Apr 2003" << doubt there's much BeOS compatibility issues there.
<OscarL>
IIRC, BeOS 5.1d0 used 3.2 for its decorators (as an static library).
<Begasus>
well, haven't even looked into the older versions, for now also ony focussing on lua5.4
<Begasus>
point is that it is confusing now ... where should the lfs and luv library really go so it's detected by the system?
<OscarL>
that's why we need someone that actually uses lua to step up :-P
<Begasus>
if they match the python stratagie they would be expected in $prefix/lib
<Begasus>
as they are C(?) libraries we should drop gcc2 support if they need to be in $prefix/lib
<OscarL>
might be similar to how Python has "C extensions" as .so under lib/python3.x/ (intended to be loaded from scripts "import c_module") ... vs libraries (the actual libpython.3x.so for example), intended to be linked by programs that need it?
<Begas_VM>
then I would have to mark gcc2 as broken too
<Begas_VM>
with luacheck .... luafilesystem >= 1.6.3 (not installed)
<Begas_VM>
either way
<Begas_VM>
no wonder the list for OS's on those packages is limited :P
<OscarL>
LUA_PATH vs LUA_CPATH...
<OscarL>
.lua files go to LUA_PATH, .so to LUA_CPATH.
<Begas_VM>
with luv it uses both $prefix/lib and $libDir
<OscarL>
"LUA_CPATH" was the one I was meaning to fix for that "vis" not finding the lua plugin on 32 bits :-)
<OscarL>
(apparently, you can "print(package.path)" and "print(package.cpath)" from a cmd:lua interpreter, to see where it thinks those are) https://github.com/TACC/Lmod/issues/112
<OscarL>
meanwhile... my attempts at fixing fork issues with Python 3.12 when nowhere, so I'll revert back to my previous solution (using spawn instead of fork for multiprocessing).
<Begas_VM>
looks like we're both stuck :)
<OscarL>
Also... dopping $jobArgs from make invocation again. otherwise, seems to still give issues related to libgcov sometimes :-( (build still works, but it is less optimized). There goes my faster builds :-(
<OscarL>
I'll report back in 35/40 minutes :-D
<OscarL>
uff, make that 45/50. had to run hp -c (and can't use hp -G as I need to re-extract a patch I've modified by hand :-P)
<Begas_VM>
heh
<Begas_VM>
ps, if you ever want to try out lazarus, set ~/config/non-packaged/bin as install path with fpcupdeluxe (seems to work there instead of keeping the default ~/fpcdeluxe
<OscarL>
cool. I'll certainly ping you when the time comes :-D
<Begas_VM>
pulls the sources for fpc and lazarus, so have to keep that in mind too :)
<OscarL>
yeah... :-(
<OscarL>
think I haven't run fpc since 2005
<Begas_VM>
well, if fpc sources are installed and setup correctly you'd only need lazarus source
<Begas_VM>
but haven't tried that
<Begas_VM>
ow, and need to patch for libiconv :P
<OscarL>
you are doing a terrible job, if you're actually trying to convince me to try it :-D
HaikuUser has joined #haiku
HaikuUser is now known as mrjones
<Begas_VM>
heh, it's logged here, so maybe someone could use it ;)
<Begas_VM>
k, luarocks installs the lfs (build in luacheck) in $prefix/lib
<mrjones>
do we have a tool that can extract metadata from mail files ('converted' eml) and write them to attributes? I am not feeling like manually fixing 28400 mails :p
* OscarL
started "hp python3.12 |& tee -a py3.12-build.log", lets see how long it takes this time.
korli has quit [Quit: Page closed]
zard has joined #haiku
<OscarL>
for writting/updating attributes, "ls /bin/*attr" will list what's avaliable, mrjones.
<OscarL>
not sure if we have something to parse the mail files themselves.
<mrjones>
mbox2mail didn't quite work for me. It created spurious mails and set the date to either 1 january 1970 or 23 may 2023 ;)
<mrjones>
.eml is a bit easier - CMD+A (wait a few minutes), right click, add-ons -> filetype -> text/x-emails and there you go. But without metadata
<OscarL>
filed a bug report already for those mbox2mail related fails? :wink: :wink:
<mrjones>
no, not yet. Also investigating scanimage failures - and some KDL-visits when copying files from my linux desktop. I'm being a little mean til my haiku
HaikuUser has joined #haiku
<mrjones>
*til -> to
<OscarL>
what about something like this: https://stackoverflow.com/a/61125183? (and then use the mentioned /bin/*attr tools to write the .eml parsed metadata as attributes?
_023621 has joined #haiku
<_023621>
.
<OscarL>
ipv6... fancy.
<mrjones>
that one could actually work out
<_023621>
Hello
<OscarL>
Hi _023621
<_023621>
Hi friends, are you using haiku os? How do you feel?
<OscarL>
Feelling pretty cold right now, otherwise, OK :-). Running Haiku on a VM at the moment (building/testing Python 3.12)
<mrjones>
Feeling kinda hot, but I am danish, so anything with sunshine is hot. Using Haiku on bare metal - first and foremost for mails and DTP.
ryder[m] has joined #haiku
<OscarL>
Are you a new, or a returning, Haiku user, _023621? Either way, welcome to our little corner of the net.
<erysdren>
hello!
<Begas_VM>
'lo erysdren :)
<OscarL>
Good morning erysdren :-)
<_023621>
You're developing Haiku? Very good! _mrjones,OscarL
<mrjones>
No, I'm only translating for now.
<OscarL>
just another user trying to contribute the little I can :-)
<_023621>
I've known about Haiku for a long time, but never installed it on my computer because of driver problems
<_023621>
I have been working on Linux flatpak from 22:00 to 4:30 yesterday
<_023621>
Still planning to give up on Linux, although it can be installed on my computer, it's a bit too difficult
<OscarL>
(running tests now... this will at least warm up the room a bit :-D)
<mrjones>
Now, try building Chromium...
<OscarL>
I don't wan't my poor PC to actually melt down, mrjones! :-P
<mrjones>
that sort of kinda actually happened to me with my previous pc build. That light grey matte PSU was darkbrown shiny when it died (during compiling something)
<zard>
Oof. I never knew that could happen
<mrjones>
it had that heat damaged metal look... i could have fried eggs on the top of the pc tower
<Begas_VM>
:)
<Begas_VM>
still not going anywhere with luacheck/luafilesystem :/
<OscarL>
time to work on something easier for a while, Begas_VM? As to get some wins and avoid frustrations?
<mrjones>
rules out BeMines
<OscarL>
how about libreoffice 6.4 (Python 2.7 and 3.7) ? :-P
<Begasus>
can get the packages to build, but don't know if they are (or even were for that matter) functional
<Begasus>
tsss :P
<Begasus>
OscarL, just revbump it in a PR and get it over with? (can always blame you then) :P
matt2 has joined #haiku
<OscarL>
and as I don't have merge powers.... I'll blame you back! Deal!
<Begasus>
lol
vdamewood has joined #haiku
matt2 has left #haiku [#haiku]
<OscarL>
definitively, we can't use "--timeout" when running Python tests. Leaves dangling processes behind after it "kills" (no very efectively) tests that timeout.
<OscarL>
now I remember why I needed "killall" and why I ended up writting /bin/pidof :-D
<OscarL>
"killall python", there... much better.
<mrjones>
hmm... now I just have to get all that stuff into file attributes
<mrjones>
when that thing has stopped parsing 28202 emails...
<mrjones>
maybe I should cut the task into smaller bites?
<mrjones>
didn't take that long after all
<OscarL>
"ValueError: time data '2023-06-21 13:30+4294966996' does not match format '%Y-%m-%d %H:%M%z'" << this might be the main cause of all the datetime.py related failures (%z == 4294966996 LOL)... will have to take a look at _strptime.py
<Begas_VM>
we as in ... ? ;)
<mrjones>
we as in plural me obviously
<Begas_VM>
heh
<zdykstra>
hah
* mrjones
forgot what I was about to do
<Begas_VM>
lol mrjones, me has that often too :)
HaikuUser2 has joined #haiku
<Begas_VM>
well, the lua* packages build and package fine so far
<OscarL>
"Fatal Python error: PyOS_AfterFork_Child: the function must be called with the GIL held" <<< this one looks nasty :-/
<mrjones>
I remember now. It was looking at the attributes of mails to make sure I get the right type
HaikuUser2 has quit []
<Begasus>
feels like merging the PR's for fife and swift_lang (to disable them) ...
<Begasus>
no one stepped up at github nor the forum
* OscarL
handles Begasus a box of matches and some gasoline.
HaikuUser2 has joined #haiku
HaikuUser2 has quit []
<Begasus>
too risky with the temperatures here :)
<OscarL>
test run complete! (with no dangling processes, nice.). 371 tests OK. 26 tests failed. 28 tests skipped.
<OscarL>
"Total duration: 13 min 38 sec." Not bad.
<OscarL>
and this time I have full logs. yay me!
xiku has joined #haiku
* mrjones
is idle: going to the grocery store for beer
<Begas_VM>
congrats OscarL! :)
<OscarL>
:-). Hope someday this will help me to actually attempt to fix the issues found by those tests, heh.
<Begas_VM>
after fidling with the config file for luarocks this works :) https://bpa.st/QRSDC
<OscarL>
So... not only LUA_PATH/LUA_CPATH, but also LUA_DIR/LUA_CDIR ? "nice" (not).
Begas_VM has quit [Quit: Vision[]: i've been blurred!]
<Begasus>
closing down, cu peeps!
Begasus has quit [Quit: Leaving]
bjorkintosh has quit [Ping timeout: 480 seconds]
HaikuUser2 has quit [Quit: Vision[]: i've been blurred!]
HaikuUser2 has joined #haiku
HaikuUser2 has quit []
zard has quit [Quit: leaving]
HaikuUser2 has joined #haiku
<pairisto[m]>
<OscarL> "pairisto: by "didn't refresh..." <- that is what I meant yeah. also @korli I tried `jam -q update-image @nightly-anyboot "<driver>tun_driver"` in /generated and I get `Error: Mounting FS failed: Bad file descriptor`
<pairisto[m]>
> <@_oftc_OscarL:matrix.org> pairisto: by "didn't refresh" you mean that "tun_driver" (and driver.o) under "generated/objects/haiku/x86_64/release/add-ons/kernel/drivers/misc/" do not change at all?
<pairisto[m]>
* that is what I meant yeah. also @korli I tried `jam -q update-image @nightly-anyboot "<driver>tun_driver"` in /generated and I get `Error: Mounting FS failed: Bad file descriptor`
HaikuUser2 has quit [Quit: Vision[]: i've been blurred!]