erysdren has quit [Quit: Konversation terminated!]
Begasus has joined #haiku
<Begasus>
g'morning peeps
<coolcoder613>
Hello
<Begasus>
Hello coolcoder613
<coolcoder613>
Could you help me with this?
<Begasus>
augiedoggie_, still around? (got a chance to look at the PR for djvu?)
<Begasus>
don''t know coolcoder613 (not a developer here) :)
OscarL has joined #haiku
<Begasus>
Hola OscarL!
<OscarL>
Hey there Begasus :-)
<Begasus>
feeling better OscarL?
<Begasus>
166 to go (already cut 12 out there) :)
<coolcoder613>
I ported Digger yesterday (digger.org)
<Begasus>
nice coolcoder613, if it works well maybe add it to haikuports?
<OscarL>
Begasus: Yeah, thanks for asking! That whole situation caught me a bit with my guard down, I think :-/
<coolcoder613>
trouble is, it needs an argument to be passed for sound
<Begasus>
been thinking about you yesterday OscarL, hope it calmed down there
<Begasus>
no sound if not passed?
<coolcoder613>
yes
<coolcoder613>
it has to be passed, even if i try and change the argument handling code it still wont work
<OscarL>
Thank you Begasus. Yeah, better on my end. Hope that poor lady's life gets better too.
<augiedoggie_>
Begasus: i'm heading to bed but i didn't see anything wrong when i looked at it earlier
<Begasus>
ok thanks augiedoggie_, should be good to go then :)
<Begasus>
coolcoder613, maybe add a startup script for it then?
<Begasus>
fresh coffee refill ...
<coolcoder613>
then i put the icon on the startup script?
<OscarL>
coolcoder613: hi, and thanks for working on those Haiku API bindings for Python! While I can't help with them directly...
<OscarL>
let me know if there's a python package/recipe you need added or updated :-D
<coolcoder613>
Can you help me decipher a debug report?
<OscarL>
BTW... noticed on the forum posts that you use Python 3.9. Just in case... 3.10 is now the default on Haiku, and has working autocomplete, and usable line editing on th REPL.
<OscarL>
coolcoder613: I can try, but no promises :-D (not too skilled here :-D)
<Begasus>
no clue here about the debug reports :/
<coolcoder613>
on the forum post about Haiku-PyAPI
<Begasus>
coolcoder613, iirc it should be possible to add an icon to the script, haven't done it myself so far though
<coolcoder613>
Begasus, AFAIK i should be able to mark the script as executable and export as BEOS:ICON from IOM
<OscarL>
coolcoder613: I *think* it looks like it is trying to send a message to some object that was already destroyed?
<coolcoder613>
or out of scope?
<Begasus>
no real help there coolcoder613 sorry :/
<OscarL>
Lifetime and ownership issues are just too complex for my poor brain :-/
<coolcoder613>
It seems to be in the middle of deleting a few things, probably when quitting SetupWindow
<Begasus>
should fiddle with that when I come to it (and write it down)
<Begasus>
jmairboeck mentioned something like that while back here on IRC, was some time ago ...
<Begasus>
roidred did something like that for python scripts iirc too
<OscarL>
the monsterz recipe adds icon to the python script, IIRC.
<coolcoder613>
only changes where to Makefile and def.h
<Begasus>
would be a small patch then :)
<OscarL>
coolcoder613: you can use "git format-patch" to extract a .patch file with your changes...
<OscarL>
then you should be able to use the original file on the SOURCE_URI for the recipe... and add your patch on the "PATCHES" variable of the recipe.
<OscarL>
And? the original tarball will be unpacked, and your patch applied to it. Should work, no Begasus?
<OscarL>
maybe I'm missing something, tends to happen :-P
<OscarL>
coolcoder613: if that's the one you imported into your git repo, then yes.
<coolcoder613>
I imported my patched version into the repo
<OscarL>
I always start with the "initial import" from the original sources, and then add my mods. Makes things easier to upstream, if needed.
<Begasus>
MacPorts/openSUSE ... they all use that archive and supply the patch on top for the build
<Begasus>
patch(es) (MacPorts has a few patches there)
<Begasus>
biab ... doggies
<OscarL>
Well... what a save! I had to "recompile" after applying some patches to my recipe, but... results are surpricingly yummy! /me steps out to eat.
HaikuUser has joined #haiku
<coolcoder613>
I made a commit that changes to the old code, now what do i do?
<OscarL>
git format-patch -p -k -1
<coolcoder613>
fatal: ambiguous argument '1': unknown revision or path not in the working tree. Use '--' to separate paths from revisions, like this: 'git <command> [<revision>...] -- [<file>...]'
<OscarL>
I've just tried command here, on one of my repos, and it works :-/
<coolcoder613>
what is it supposed to do?
<OscarL>
the -1 is so it only extracts a patch for the last commit.
<coolcoder613>
i just changed it back again
mr_lou has quit [Quit: Leaving]
<OscarL>
ah... That "go back to original sources" was not what I expected :-D
<OscarL>
I would just start afresh, with the first commit being just the original sources, and then my changes.
<OscarL>
It should be possible to do it from your current state, but I'm not that proficient at git, and less so to explain via IRC, sorry!
<coolcoder613>
Well, i can make a *new* repository from the original sources.
<coolcoder613>
I dont really use command-line git, i am handling the git stuff with github desktop on my mac via sshfs
<Begasus>
re
<Begasus>
when I want to upstream patches I fork that one, clone it localy, create a new branch, make changes, create commit and push to origin (and then file a PR)
<coolcoder613>
OK, now i have a local repo with original digger source, what do i do now?
<Begasus>
or in this case, with the original code in "original", clone, create branch, make changes, git diff, git commit -a ...
erysdren has joined #haiku
<OscarL>
coolcoder613: add your needed changes, either as a single commit, or multiple, and then use that "git format-patch -p -k" command (making sure the current directory is in the working tree)
<Begasus>
if your code only changes for those 2 lines the diff should a look a bit like this? https://bpa.st/4VCA
<coolcoder613>
What about git diff
<OscarL>
that commands needs to know what to extract... either you pass it -"n" (with "n" being the number of commits "from the HEAD" you want to extract)...
<coolcoder613>
OBJS += # insert here the names of the files which contains various missing functions like strup() on Linux and FreeBSD
<coolcoder613>
-CFLAGS += -DFooOS # insert here additional compiler flags which required to find include files, trigger os-specific compiler behaviour etc.
<coolcoder613>
-LIBS += # insert here libs required to compile like zlib, SDL etc
<coolcoder613>
+CFLAGS += -DHAIKU `sdl-config --cflags` # insert here additional compiler flags which required to find include files, trigger os-specific compiler behaviour etc.
<coolcoder613>
+LIBS += `sdl-config --libs` -lz # insert here libs required to compile like zlib, SDL etc
<coolcoder613>
ESUFFIX = # insert here suffix of the executable on your platform if any (like ".exe" on Win32)
<coolcoder613>
endif
<coolcoder613>
<coolcoder613>
diff --git a/def.h b/def.h
<coolcoder613>
index fd1fd6c..5894844 100644
<coolcoder613>
--- a/def.h
<coolcoder613>
+++ b/def.h
<coolcoder613>
@@ -10,7 +10,7 @@
<coolcoder613>
#endif
<coolcoder613>
#endif
<coolcoder613>
<coolcoder613>
-#if defined FREEBSD || defined LINUX || defined YOUR_UNIX_LIKE_ARCH_GOING_HERE
<coolcoder613>
+#if defined FREEBSD || defined LINUX || defined HAIKU
<coolcoder613>
#endif
<coolcoder613>
#define UNIX
<coolcoder613>
Is that the right format?
<OscarL>
and now you know to not do that on IRC...
<Begasus>
the same as the one I just linked (ps, use paste online) :)
<Begasus>
DjView also done on 64bit buildmaster :)
<OscarL>
should work, coolcoder613, difference between that format and the one from format-patch, is that the diff do not include your commit message, and stuff.
* OscarL
looks at that URL. heh.
<Begasus>
that's what "git commit -m/-a ..." does OscarL :)
<coolcoder613>
How can i create a recipe? i want a template or a tutorial
<Begasus>
but I guess format-patch "actually" creates a *.patch file?
<OscarL>
Begasus: I mean... if you already done that... format-patch includes that on the output, so you can just use it that way, yeah
<Begasus>
coolcoder613, there are some generic recipes in the haikuporter directory
<OscarL>
Begasus: remember to use -p -k with format-patch, so it avoids adding "[PATCH n/m]" on the "Subject:" line.
<Begasus>
not seeing any output here OscarL, so I don't know what has been done :P
<OscarL>
the "rdef" file is what's used to add flags/icon/supported filetypes to a binary (after getting converted to an .rsrc file IIRC)
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
<OscarL>
the "rdef.in" is just a template that gets filled by the .recipe, turning it into ".rdef" (still a text file), and that gets turned into proper .rsrc binary resources that get added to the binary.
HaikuUser has joined #haiku
<Begasus>
haikuporter magic (python) :)
<OscarL>
and looots of bash and sed :-P
<Begasus>
:)
<coolcoder613>
how do i do i do multiple copyrights
<Begasus>
undefined reference to `major' ... that's a first
<Begasus>
OscarL "lgrep major" doesn't really help here :)
<OscarL>
>>> <Begasus> but when do you need it?.... when, for example, you want to extract a patch from a particular commit (or commits) from one repo, to apply to another... you can use `git format-patch -p -k -<n> <rev>`
<Begasus>
Ah probably will need parts of the patch from korli
<Begasus>
long time ago since I fiddled with "real" patch files OscarL :)
<OscarL>
Begasus: too many matches?
<coolcoder613>
do i just paste the git diff output to the *.patchset file?
<Begasus>
yeah, and not a good one
<Begasus>
coolcoder613, did you push your changes to your repo? (eg and using that) or are you using the original archive?
<OscarL>
Begasus: the unreference only says "major" ? no other hints from were to look? BSD/GNU, etc?
<coolcoder613>
I am using the original tarball as source
<OscarL>
notice the "cp haikuports-sample.conf /boot/home/config/settings/haikuports.conf # Copy the config file" step
<OscarL>
the first time it all looks just crazy... but it gets better with time :-D
<OscarL>
(or it might just be the Stockholm syndrome kicking in here :-P)
<Begasus>
SUMMARY can't start with the port name
<Begasus>
DESCRIPTION has a double qoute to much (lines must not exceed 100 characters)
<Begasus>
no dot at the end in SUMMARY
<Begasus>
Warning: POLICY WARNING: "/packaging/digger/apps/Digger" needs library "libSDL-1.2.so.0", but the package doesn't seem to declare that as a requirement
<Begasus>
Warning: POLICY WARNING: "/packaging/digger/apps/Digger" needs library "libz.so.1", but the package doesn't seem to declare that as a requirement
<Begasus>
w
<Begasus>
coolcoder613, are you on 64bit?
<coolcoder613>
Yes, but i made the code changes on 32 bit
<Begasus>
ah, did you build that with gcc2 there?
<OscarL>
Begasus: dovecot's major() is not a function, but a macro.
<coolcoder613>
no
<coolcoder613>
Error: digger not found in repository
<Begasus>
cleaned those long ago, would rather have it stay like that :)
mmu_man has joined #haiku
<Begasus>
using lib:libsdl turns into: Warning: POLICY WARNING: "/packaging/digger/apps/Digger" needs library "libSDL-1.2.so.0", but the package doesn't seem to declare that as a requirement
<Begasus>
that cleaning worked well :P
<Begasus>
problem 1: nothing provides lib:libsdl needed by digger-20020314-1
<OscarL>
erysdren: is SOURCE_DIR="quakespasm-quakespasm-$portVersion" ok? Double "quakespasm"?
<erysdren>
yeah, that's how it is in the source tar.gz
<erysdren>
for some odd reason
<OscarL>
ok. just making sure.
<erysdren>
thanks Begasus :D
<Begasus>
thanks :)
* OscarL
debates... a round of Quake... or a round of Python 3.12 pacthset... mmm
<Begasus>
got around to push some patches upstream for that OscarL? ;)
<OscarL>
Begasus: need to streamline them first. We have, for example, a couple of commits that should be squashed into one...
<OscarL>
after separating some bits from the bigger one into a new one... stuff like that.
<OscarL>
my lack of focus, and my need to double-triple check things there are making me slower than usual :-D
<erysdren>
where would a ported compiler go? dev-lang?
<Begasus>
try with some that are basics OscarL, no hustle for upstream, should narrow down patching before the official release?
<OscarL>
Also, my idea is to just stream line them as much as I can, but let others do the upstreaming... dealing with python.org folks is out of my league.
<Begasus>
haven't got anything back on the upstream patch for fpc :/
<OscarL>
Begasus: yeah, current patchset is almost 100% as good as our previous ones (for our use in Haiku at least). Need to see if that weird build issue you had still reproduces, for example.
<Begasus>
which one? :P
xet7 has joined #haiku
<OscarL>
you had some "random" build failure when building 3.12.
<OscarL>
that I've only seen once in the past (and I think that was for 3.11), but couldn't reproduce.
<Begasus>
I right, forgot about that
<OscarL>
one complaining about not finding libpython3.12.so, IIRC
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
KapiX has quit [Quit: KapiX]
HaikuUser has joined #haiku
HaikuUser has quit []
<OscarL>
erysdren: dev-lang sounds reasonable, which compiler? (might look up where Gentoo places it)
<OscarL>
Alright! I came up with a bash function to do inrecipe for each "cmd:" a recipe provides. Shows that yeah.. no .recipes uses nothing from csound :-D
<OscarL>
time to add that to my "bashrc" :-D
hightower2 has quit [Ping timeout: 480 seconds]
<Begasus>
csound should be quite active iirc, had some talks to it's maintainer iirc
<Begasus>
or was that something else ... ;)
<OscarL>
no idea, just wanted to make sure if something actually used it in-tree before attempting to update it (and remove the old recipes)
hightower2 has joined #haiku
<Begasus>
+1
<erysdren>
bleck, fteqw keeps saying opus.h not found
<erysdren>
even though the dependency is set
<OscarL>
mmm, can we nest functions in bash? goin to find out soon enough :-D
<OscarL>
heh, that worked :-D
<erysdren>
the annoying thing about fteqw is that it insists on building and linking a lot of libs statically
<erysdren>
for portability
<Begasus>
nice, DigeKam detects when exiftool is installed :)
<Begasus>
better then to bundle them as shared erysdren
<puck_>
or, hrm, not even sure that should matter, virtual APICs should allow virtualising arbitrary cluster IDs
Forza has joined #haiku
<Coldfirex>
The physical host has dual 16-core, 32 thread cpus
<andreasdr[m]>
Hi there
<andreasdr[m]>
Puck: I saw your commit. Interesting.
<puck_>
yeah, i got hyperfocused onto the scheduler and found some very subtle bugs
<puck_>
though, the ICI one is less subtle
<puck_>
(my guess is that's why the scheduler sometimes just hangs, and recovers if you enter and exit kdl)
<puck_>
uh, not the scheduler, but
<Coldfirex>
well fighting the mouse now. It jumps from side to side. I think I had vmware tools installed before to fix that, but HaikuDepot doesnt show it. Only lists 191 software packages...
<OscarL>
(just adding a license, we can do the rest if your too busy :-D)
<puck_>
there
<OscarL>
awesomo, thanks!
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
<Coldfirex>
ah, anyboot is missing tons of packages/updates. Hopefully these will get it back to normal
<puck_>
Coldfirex: did you use the prebuilt iso?
<puck_>
you want to avoid updating haiku.hpkg. as that'll undo the kernel patch :p
<Coldfirex>
I did
<Coldfirex>
uhh, how do I prevent that but still be able to use my mouse? :)
<puck_>
uh, sec
<Coldfirex>
software updater is running, and of course there is no cancel button lol
<puck_>
if you reboot, hit .. etiher shift or space, i forget, and revert to older activated packages
<Coldfirex>
Rebooted. It hadnt started applying yet
<puck_>
Coldfirex: on the iso, i can just run `pkgman install vmware_addons` and it picks up only that hpkg
tqh has quit [Remote host closed the connection]
<puck_>
might've also been haikudepot being a bit slow
<Coldfirex>
Either I just switched to 64vcpu. LEts see if it even stays up on its own
hightower2 has quit [Read error: Connection reset by peer]
bbjimmy has joined #haiku
hightower2 has joined #haiku
<Coldfirex>
Its still up, but I cant do much w/o mouse. Want me to drop to kdl or just let it sit?
<puck_>
Coldfirex: hit alt+super+t, then `pkgman add vmware_addons`
<puck_>
uh, install*
<puck_>
but other than that, if it boots up and doesn't seem to hang, that seems good
<Coldfirex>
much better!
<puck_>
i mean, ICIs are no longer being sent to the wrong core, i'd expect it to behave a lot better :p
<Coldfirex>
Seems to be!
<Coldfirex>
The mouse clicks dont always register but maybe thats an issue with the addons
<Coldfirex>
think we at least have an uptime record
<puck_>
heh. i mean, 64 emulated cores on a 32-thread machine under qemu also work pretty reliably here
<puck_>
though i think having actually 64 threads changes the behavior a bit
<puck_>
(and ESXi is weird with its cluster IDs)
<puck_>
Coldfirex: is this revision 2 or 4 of the CL?
<Coldfirex>
CL?
<Coldfirex>
hrev57305_6919_2
<puck_>
yeah, that's revision 2 of CL 6919
<puck_>
(changelist) that should be fine, i think; i changed it a tiny bit in revision 4, but it shouldn't change how it works for you
<Coldfirex>
OK, I just used the link in the ticket
<puck_>
yeah, that link is no longer working because rev4 has been built :p
<Coldfirex>
Makes sense. I can checkout rev4 but maybe not until this weekend
<Coldfirex>
but we are still running
<puck_>
yeah, no worries, this looks like it'll stay working
doppler has quit [Ping timeout: 480 seconds]
<Coldfirex>
Appreciate the work on this!
<Coldfirex>
Im back to conference, later!
* OscarL
played with disabling cores on VBox via ProcessController (beta4 64 bits)... Ended up with a mostly frozen system :-D
doppler has joined #haiku
<BlueSky76>
OscarL: You can disable cores with ProcessController? Cool, didn´t know that :-)
<BlueSky76>
Let me try ...
<BlueSky76>
Disabling one core works for me here on my VBox system. 64bit System, hrev57177
HaikuUser has joined #haiku
<BlueSky76>
Now I´m curious what happens if I try to disable the remaing second core :-)
HaikuUser has quit []
HaikuUser has joined #haiku
<BlueSky76>
Oh, nice. You actually get a warning message: "This is the last active processor. You can´t disable that." and a button with "Not funny" :-)
Coldfirex has quit [Remote host closed the connection]
<HaikuUser>
Hi
HaikuUser has quit []
<BlueSky76>
Enabling the 1st core again also works flawlessly
<OscarL>
BlueSky76: Hi! I did some combination of disablen cores 4, then 3 (out of 4), and setting unsetting "Power Saving"...
abhinavmir has joined #haiku
<OscarL>
ended up with a system that onlt responded to Alt+SysRq+D :-D
Anarchos has joined #haiku
<BlueSky76>
Oh, I see. Maybe had something to do with the power saving
<OscarL>
and then reboot, because I can't do much on KDL :-D
matt3 has joined #haiku
<OscarL>
Tried that in two different occasions, both same results. I was just toying around that after reading https://dev.haiku-os.org/ticket/17233
<BlueSky76>
I only know the "continue" cand "reboot" commands in KDL. No luck with "continue" most of the times
<puck_>
hmmm
<OscarL>
BlueSky76: cont works great for some spuriours trips into KDL due to the virtualbox additions :-D
<Begasus>
re
<Anarchos>
hello
<puck_>
OscarL: there's one possibility i can come up with for rare freezes, at least, i think. there's probably more, disabling cores is not really something done often
<Begasus>
hi Anarchos
<OscarL>
BlueSky76: the clipboard integration in VBox is particularly buggy :-D
<Anarchos>
BlueSky76 i often use «es» before continue
<Anarchos>
i mean
<puck_>
OscarL: (any threads locked to the core being disabled will likely never be scheduled again; i should look into that)
<Anarchos>
«sus»
<OscarL>
puck_: just in case, this is on beta4 64 bits (on a Phenom II X4 host)
<puck_>
OscarL: what's the highest "logical apic id" you seee in the syslog?
<Anarchos>
puck_ ny threads locked to the core being disabled will likely never be scheduled again : i wonder if it is a bug, or a feature....
<puck_>
Anarchos: unexpected interaction; this is quite rare (would likely only happen during memory map flushes)
<BlueSky76>
Anarchos: What is the "sus" command for?
<OscarL>
puck_: I can try to "freeze" it again if you want (after I save/sync)... and if you guide me a little on what to look for at KDL.
<OscarL>
So... no "make clean" for CMake stuff right? just delete the "build" dir instead? (assuming "cmake -S . -B build")
jojo_autoboy[m] has joined #haiku
<Begasus>
"make -C build clean"?
<Begasus>
or "rm -rf build" yes ;)
<OscarL>
I mean for the things that CMakes detects, yeah.
<Begasus>
CmakeCache
<Begasus>
CMakeCache.txt
<Begasus>
remove that and build again
<OscarL>
thanks. wasn't it anyway... not sure why it luaCsnd6.so is missing :-D
Coldfirex has joined #haiku
<OscarL>
"BUILD_LUA_INTERFACE:BOOL=ON" on CMakeCache.txt... oh well... I'll come back to this later on.
<Coldfirex>
@puck: VM still up. I notcied that one vcpu tends to hang around 55% usage constantly just sitting there with only activity monitor open. Process controller just shows kernel_team. Not sure if relevant
<OscarL>
"LUA_LIBRARY="LUA_LIBRARY-NOTFOUND", so disabling BUILD_LUA_INTERFACE" on the console, LOL.
<Coldfirex>
@puck_ *
<puck_>
Coldfirex: hmmm, unlikely to be relevant, but the fact it's still running seems good
<Coldfirex>
Ill leave it up and running. I'll get an email alert if it KDLs and consumes all 64 production cores
<puck_>
hahahah
<puck_>
i don't expect KDLs
nerdistmonk has quit [Quit: Leaving]
<Begasus>
that's one of the major things in csound I thought OscarL?
<puck_>
Coldfirex: tbf, the primary quirk i needed to test was the fact that your virtual machine has more cluster IDs than necessary, and i .. don't entirely know how to make qemu do this
<OscarL>
Begasus: "-DLUA_LIBRARY=$libDir/liblua.so.5.1 \" seems to have fixed it, but it ain't pretty :-D
<OscarL>
nah... failed further down.
<Begasus>
ah, so it can't detect the path without it :/
<puck_>
Coldfirex: so it looks like this likely solves the primary issue
<Coldfirex>
Not sure about that, but it seems to have done the trick! I wonder if any other platforms were affected by this (hyper-v, proxmox, etc)
<Begasus>
what was I doing?
<Begasus>
ah! :)
<Coldfirex>
I might be able to test on Nutanix which uses kvm\qemu
<puck_>
Coldfirex: other platforms, including bare metal, are in fact affected
<puck_>
Coldfirex: but e.g. on qemu, only if you have >16 vCPUs
ClaudioM has joined #haiku
<Coldfirex>
ah interesting
Coldfirex has quit [Remote host closed the connection]
extrowerk has quit [Read error: Connection reset by peer]
extrowerk has joined #haiku
hightower2 has joined #haiku
hightower3 has joined #haiku
hightower2 has quit [Read error: Connection reset by peer]
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
nephele has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
matt3 has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
matt3 has left #haiku [#haiku]
dqk_ has quit [Quit: leaving]
hightower3 has quit [Read error: Connection reset by peer]
matt3 has joined #haiku
matt3 has left #haiku [#haiku]
Begasus_32 has quit [Quit: Vision[]: Gone to the dogs!]