Yoke has quit [Quit: Vision[]: i've been blurred!]
HaikuUser has joined #haiku
pbsds is now known as Guest13560
pbsds has joined #haiku
Guest13560 has quit [Ping timeout: 480 seconds]
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
HaikuUser has joined #haiku
HaikuUser2 has joined #haiku
HaikuUser has quit []
HaikuUser2 has quit []
vdamewood 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]
Nasina has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
Nasina has joined #haiku
KapiX has joined #haiku
<brickviking[m]>
I've struck a small problem with one of the recipes I cobbled together. I created a .hpkg that differs (in version) from the latest release. Yet when I check back in HaikuDepot, it thinks that the original version got installed, not my currently-installed version.
<brickviking[m]>
The latest release version of fossil is fossil-2.25, I installed fossil-20250411160905 instead. Yet HaikuDepot says 2.25 got installed, as that's the version showing in the Version column.
<brickviking[m]>
When I run the command in the shell (fossil version) it shows the correct version, not the fossil-2.25 version from the normal release package. I'm not sure if you need any more information than that. I'm wondering how best to solve this.
xgpt has quit [Quit: Ping timeout (120 seconds)]
xgpt has joined #haiku
mmu_man has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
pvalue has joined #haiku
OscarL has joined #haiku
<OscarL>
brickviking[m]: which version do you see if you run "pkgman search -iD fossil" on Terminal?
<brickviking[m]>
thankfully, the version I've installed:
<brickviking[m]>
That's the version broken up into its respective fields, I'm guessing.
Nasina has joined #haiku
<OscarL>
"-i" means, only shows installed packages, if you remove it, you should see both versions (the "<system>" one from the installed package, and one for "HaikuPorts", the one on the repos).
Nasina has quit [Read error: Connection reset by peer]
metalist has quit [Ping timeout: 480 seconds]
metalist has joined #haiku
<brickviking[m]>
What's the -D for then? I'm getting bumkis out of the help page, with "pkgman search ..." only showing -a/--all as an available parameter
<OscarL>
"details"
<OscarL>
"pkgman search --help"
<brickviking[m]>
heh. So, any ideas about why HaikuDepot is "getting the version wrong", as such?
<brickviking[m]>
Ah, so you do what fossil does... heh. Figures.
<OscarL>
I think HaikuDepot only shows packages from the repo, not local ones.
<brickviking[m]>
pffft. Figures.
<OscarL>
does it shows an "Uninstall fossil" button if you select that 2.25 package?
<brickviking[m]>
... which means I'd have had to point it at the repo on my hard disk for it to search in there.
<brickviking[m]>
Yes, and that works.
<brickviking[m]>
i.e. typing "fossil" at the commandline basically says "/bin/fossil: no such file" (paraphrased)
<OscarL>
Guess you could/should file a ticket about it.
<brickviking[m]>
At the git repo for HaikuPorts, or somewhere different?
<brickviking[m]>
HD seems to be getting confused by the fact that fossil is installed, though the version installed differs.
<brickviking[m]>
Trac. I hadn't seen that for quite some time; it's interesting that it's still being used.
<OscarL>
hrev58811 on 32 bits VMware doesn't wants to boot :-/
<OscarL>
reboots itself as soon as it reaches the disks icon. Trying to boot to previous state hangs... time to manually replace packages.
ChaiTRex has quit [Quit: ChaiTRex]
<OscarL>
that worked, now back to beta5 on 32 bits.
ChaiTRex has joined #haiku
deneel has quit [Ping timeout: 480 seconds]
mmu_man has quit [Ping timeout: 480 seconds]
Nasina has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
<brickviking[m]>
The correct version is definitely installed at /system/packages/ too. I'm just paging through HaikuDepot-related bugs now.
<brickviking[m]>
Well, I'm not exactly finding anything directly related, though 11 years ago, someone suggested removing an older package before installing a newer package. (issue #12090)
<brickviking[m]>
I'm also wondering if (when using hp to compile a package) that program fills in $portVersion from the fed-in package name, i.e. where I use: haikuporter fossil-20250411160905, then $portVersion will then be 20250411160905. Is that correct?
Viking667 has joined #haiku
wicknix_ has quit [Ping timeout: 480 seconds]
pvalue has quit [Quit: Vision[]: i've been blurred!]
<B2IA>
(UnrealNeil) Butler catsup
<OscarL>
brickviking[m]: yes. if a recipe is named foobar-1.2.3.recipe, then $portVersion will be 1.2.3
<brickviking[m]>
Huh. That'll be why the packages don't match their initial shasums.
janking has joined #haiku
<brickviking[m]>
That, and the Haiku team are using sha256sum, instead of sha3-256
<janking>
Good morning to you all :)
<brickviking[m]>
Good morftervening to you too.
<janking>
:)
<B2IA>
(UnrealNeil) Good afternoon from Australia
<OscarL>
brickviking[m]: not sure what you mean by "initial checksum". hp computes the sha256 over the whole tarball, and compares it to the value present on the .recipe.
<OscarL>
hey there janking, and UnrealNeil.
<brickviking[m]>
B2IA: 14:43, 14:13, or 12:43 where you are?
<OscarL>
brickviking[m]: "B2IA" is the name of the bridgh between IRC and beshare, not a nick.
<brickviking[m]>
Ahhhh, right. So it's UnrealNeil I needed to have asked that time question to.
<B2IA>
(UnrealNeil) 14:45 In Cairns
<brickviking[m]>
Is it still warm up there, or are things finally cooling down?
<B2IA>
(UnrealNeil) Cairns.... Tropical... It'll be warm until August. then there will be 3 weeks, when it only gets to 20c and then back to warm again...
<janking>
06:50 Sweden
<B2IA>
(UnrealNeil) ...our two seasons and Wet and Dry. We don't have 4 like most
<brickviking[m]>
OscarL: We struck a small problem with that sha256sum. fossil authors create the tarballs from the fed-in filename. The directory component inside the tarball includes the name the tarball is downloaded with; as a result, every time we change the downloaded tarball name, the sha256sum changes.
<brickviking[m]>
It affects the recipe, as each time you change the name of the recipe, even if you use the same hashID, the sha256sum will be entirely dependent upon the name of the recipe. Ugh.
<OscarL>
brickviking[m]: I don't see the problem. tarball name changes in-sync with the version. if you ask fossil for the same file name and hash combination... the calculated sha256 remains the same.
<brickviking[m]>
(in this context, the hashID is the "tag" value inside the recipe file.
<B2IA>
(UnrealNeil) hash, context, tags, recipes... this is a discussion about making fortune cookies....
<OscarL>
brickviking[m]: it evidently has been working fine for all the updates the fossil recipe got over the years, so no idea why you think it is an issue.
<brickviking[m]>
And because the pathname is different, the calculated sha256sum differs, even for exactly the same filetree content.
<OscarL>
(for your example).. yes, and that doesn't matters in the context of the recipe, as it only ever uses one filename and not the other.
<brickviking[m]>
Ugh.
<B2IA>
(UnrealNeil) Ugh.
<brickviking[m]>
It does make it rather awkward if I want to create a recipe called fossil-current.recipe. the sha256sum will never be correct until I've fetched the initial tarball as fossil-src-current.tar.gz
<brickviking[m]>
When that recipe updates, the filename won't change unless I change the short hash inside the above link. (in this case, the 41cd8d51ba7ce1a5 piece.
<OscarL>
that's just not how hp works.
<brickviking[m]>
So I'm finding out.
<OscarL>
you can't have a fossil-current.recipe and automagically make it build the latest tag.
<brickviking[m]>
And that's exactly what I'm trying to do.
<OscarL>
won't work.
<brickviking[m]>
In essence, the way I'd have to do it would be as follows: remove the existing fossil-current.hpkg, create a new version of fossil-current with a different $tag value, get hp to recreate the new package, and install that new package.
<OscarL>
if it is for you personal use... remember that you don't need to package things, you can just "make && make install" into the "non-packaged" hierarchies.
<brickviking[m]>
Yes, I'm aware of that.
vezhlys has quit [Remote host closed the connection]
<B2IA>
(UnrealNeil) I have a windows PowerShell script, that does most the hard work of making a Haiku bootable UEFI install disk, except the formating og the BeOS Partition... and I'm hoping to do the whole thing instead, in Haiku BaSH...
<B2IA>
(UnrealNeil) Does DriveSetup have any command line switches, for creating Partitions and Initialsing them? or am I going to have to play with 'Hey' verbs?
<brickviking[m]>
As it is, not only do I have to remove the fossil-20250411 package, I also have to change the name of the recipe, as well as the $tag inside the recipe and the sha256sum
<OscarL>
brickviking[m]: I guess that you could get away without changing the recipe name.... but you will at the very least have to increment the REVISION value each time you change the hash (and adjust the SHA256).
<Yoke>
Does anyone know how well Abaddon works? the discord client
goldentooth has joined #haiku
Babaj has joined #haiku
jmairboeck has joined #haiku
Babaj has quit [Quit: Leaving]
Babaj has joined #haiku
Babaj has quit [Remote host closed the connection]
Babaj has joined #haiku
Aedil has quit [Quit: leaving]
goldentooth has left #haiku [#haiku]
<gordonjcp>
Yoke: I don't recommend it
<gordonjcp>
Yoke: as a bit of software it's probably okay, but then you'll be on Discord
<Viking667>
Aside from it being "commercial", what's everyone got against the Discord platform anyhow?
Nasina has quit [Read error: Connection reset by peer]
Babaj has quit [Quit: Leaving]
<gordonjcp>
Viking667: it's full of arseholes
Nasina has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
<Viking667>
huh. Just like X, facebook, IRC and a number of other social platforms. That's no news to me - "arseholes" seems to be a human contagion.
<|cos|>
Viking667: They require to know your legal identity, by blocking account creation on phone number verification, in order to be able to target ads. I find all of that unacceptable for what they have to offer in return.
<Viking667>
I refuse to give them a phone number. It's beside the fact I don't have a phone number they can use anyhow.
<Viking667>
And let's face it, a phone number is not a legal identity in any jurisdiction I know about.
<|cos|>
Viking667: …and you still got an acount? Then you provided them a tracking cookie which effectively means they already had your identity bound to a phone number?
<Viking667>
What phone number? I never gave them one. For the servers that require a phone number, I decided not to join them.
<gordonjcp>
Viking667: yes, but different platforms have different arseholes
<gordonjcp>
Viking667: I'm not on Twitter because it's full of Nazi arseholes
<Viking667>
gordonjcp: mmm hmm.
<|cos|>
Viking667: It's a unique hard to change identifier, and with the repressive trend of lately there are few countries where anonymous phone numbers are still allowed.
<gordonjcp>
I'm not on for example Mastodon because that's full of gay neonazi furry arseholes
<Viking667>
sigh.
<|cos|>
Viking667: Are there other servers than discord.com?
<Viking667>
not referring to that. Each "interest group" (such as Smalltalk, Emacs, Minecraft) is identified on the platform as "a server".
<Viking667>
It's that to which I'm referring when I talk about servers.
<|cos|>
Viking667: …but since teh "servers" require an account with the actual server, which is blocking account creation on phone number verification that's not quite relevant.
<|cos|>
Viking667: Anyways. You've got your answer, with explanation. Debating that is a bit too off-topic for me.
<Viking667>
I had no trouble creating accounts. At no time was I asked for a phone number until I attempted to join certain very specific interest groups. Most other interest groups are fine with anyone joining up.
<|cos|>
Viking667: You might wish to challenge your assumption that Discord is your actual friend, re-read chat history and accept reality.
<Viking667>
I never stated "Discord is my friend". However, I will say that on that platform, I have not been socially challenged. Perhaps it's the lack of servers I'm part of. I tend to restrict what I talk about.
<Viking667>
I've found the same for IRC, Skype, Google Talk (when it was a thing) and Zoom.
Nasina has joined #haiku
<Molnija>
oh good grief.
Nasina has quit [Read error: Connection reset by peer]
<Viking667>
Yeah. Call me sheltered. That way I avoid most issues.
<Viking667>
Hi Molnija.
<|cos|>
Viking667: I think you're missing the point, but it's all off-topic. Good luck finding an open social arena without human behaviour!
nephele has joined #haiku
Nasina has joined #haiku
<Molnija>
|cos|: discord's server confusion is interasting. idk why they do that
HaikuUser has joined #haiku
HaikuUser has quit []
<nephele>
Discord calls these "guilds" internally, so you should probably to if not talking to people in the "in" group
<Viking667>
? something's .... "you should probably to if not talking to people in the in group"?
<nephele>
I don't understand your question
<Viking667>
I didn't entirely understand the second half of your question either.
<|cos|>
nephele: s/to/too/, right? i guess that's the confusion
<Viking667>
could be, I'm not sure - I thought I'd ask.
<nephele>
right, should be too
<Viking667>
Ah. Now it's clear.
<Viking667>
"do" didn't feel right somehow. "too" makes more sense.
<Viking667>
Anyhow, I'm headed back to Feist's "Magician: Apprentice".
Viking667 has quit [Quit: *croak*]
janking has joined #haiku
janking has quit []
<nephele>
Begasus[m]: any progress on that quaternion fix? :g
<nekobot>
• Begasus (80977d0a): hippo, revbump for poppler changes, keep to poppler24 for now (#12149)
nipos has left #haiku [Disconnected: Received SIGTERM]
nipos has joined #haiku
pol has joined #haiku
HaikuUser3 has joined #haiku
HaikuUser3 has quit []
lubo76kubuntu has joined #haiku
vezhlys has joined #haiku
<OscarL-32>
darn Python 3.11 still has issues when doing parallel optimized builds. back to "make -j 1".
lubo76kubuntu has quit [Remote host closed the connection]
HaikuUser has quit [Ping timeout: 480 seconds]
AlienSoldier has joined #haiku
HaikuUser2 has quit [Ping timeout: 480 seconds]
<OscarL-32>
silly 3.11 has missing ffi symbols that cause the _ctype module to be disabled (at least on 32 bits nightly). let's try disabling the use of system wide libffi.
<Begasus[m]>
whoa!! :)
<Begasus[m]>
ps, just saw the email :)
* OscarL-32
wishes he had a couple n100 based mini PC to run a bunch of builds at once (beta5 vs nightly, 32 vs 64 bits, custom build options, etc, etc).
<Begasus[m]>
cluster builds? ;)
<OscarL-32>
fatal error: ffi.h: No such file or directory. Good thing I tried an unoptimized, parallel build to test that.
<OscarL-32>
meh... 3.11 is an odd ball anyway... if 3.12 and 3.13 work OK.... I won't look deeper into this _ctype thing on 3.11.
<scanty>
i'm in the process of determing if it is a haiku problem, or pianobar itself.
janking has joined #haiku
<andreaallegri>
OpenBeOS = in italian sounds like : to be open and operative
metalist has quit [Read error: Connection reset by peer]
metalist has joined #haiku
<andreaallegri>
be operational
<andreaallegri>
to be operational
<OscarL-32>
Python 3.12 creates the _ctype extension without problems on 32 bits... will just blame 3.11 for being odd as usual.
pol has quit [Quit: ~(]|[)~]
<andreaallegri>
OpenBeOS in italian sounds like : to be open minded and operational
janking has quit [Quit: Vision[]: i've been blurred!]
<gordonjcp>
OscarL-32: how hard would it be to set up a build server?
<andreaallegri>
peace
andreaallegri has left #haiku [#haiku]
<OscarL-32>
gordonjcp no idea, really. haven't used haikuporter in "buildmaster" mode yet.
<gordonjcp>
OscarL-32: I have a Hetzner box with a fairly chunky CPU, 64GB of RAM, and a fucktonne of disk space
<gordonjcp>
OscarL-32: since it runs half a dozen docker containers as web servers and maybe four or five Windows XP VMs (funny story behind those) it's basically idle
<OscarL-32>
If I learn enough about how to setup things (on the haikuporter side of things)... I might get back to you :-)
<gordonjcp>
:-)
<OscarL-32>
build times on these Pythons can vary *wildly* depending on build settings. Can go from < 2 minutes (just "./configure && make") to about an hour (fullly optimized, LTO-enabled .hpkg).
<OscarL-32>
"test_datetime passed", same with 3.10... and somehow that one ALWAYS fails on 64 bits.
<scanty>
OscarL-32: lto is really slow.
<scanty>
in general
<OscarL-32>
yeah :-/
<scanty>
it takes 7 seconds here to link a 800KB binary
<gordonjcp>
what holds it up?
<OscarL-32>
linker being slow as molasses, IIRC.
<gordonjcp>
is it just inefficient?
<OscarL-32>
think so, yea.
<scanty>
lto just takes a while.
<scanty>
it's the last stage of optimisation.
<scanty>
you can build without it no problems
<OscarL-32>
here I'm using -O3 + PGO + LTO, so triple whammy.
<scanty>
what's PGO
<OscarL-32>
profile guided optimizations.
<scanty>
how does that work?
<OscarL-32>
build version A, run version A against a bunch of tests, collecting profile data...
<OscarL-32>
next phase compile things again, giving the compiler the profile data, so it can perfom better optimizations.
<scanty>
wow that's cool. i never even heard of that ;-)
<OscarL-32>
the "/bin/package create" steop is also very slow (and single-threaded) :-(
<scanty>
hrm, i'm playing with compiler flags now.
<OscarL-32>
will attempt to run the test-suite now... let's see if it sends me to KDL like the other day :-D
HaikuUser has joined #haiku
HaikuUser has quit []
<scanty>
hm cleaned up my makefile a little.
CompanionCube has joined #haiku
<scanty>
had a -gdwarf-2 -gstrict-dwarf in CXXFLAGS
<scanty>
don't know why
<scanty>
don't remember why, rather.
<OscarL-32>
older Haiku Debugger had issues using newer dwarf versions. Maybe that's why?
* OscarL-32
disables yet another test that hangs. Sigh.
<scanty>
perhaps
<scanty>
hmm, didn't need -m64 either.
nephele has quit [Quit: Vision[]: i've been blurred!]
lubo76kubuntu has joined #haiku
lubo76kubuntu has quit [Remote host closed the connection]
<OscarL-32>
some of Python test cases are very fragile. Who in their right mind writes an assert based on the wording used in an OS level error message?
<scanty>
what do you mean?
<OscarL-32>
test expects that the error message matches a pattern: "missing.*?.so"
Babaj has joined #haiku
<OscarL-32>
where in Haiku, the error message is just: "No such file or directory".
<gordonjcp>
OscarL-32: jesus
<gordonjcp>
syscalls motherfucker do you speak it
jmairboeck has quit [Quit: Konversation terminated!]
Nasina has quit [Read error: Connection reset by peer]
<scanty>
ah
Nasina has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
Nasina has joined #haiku
Nasina has quit [Remote host closed the connection]
<nekobot>
• pulkomandy (fc90b072): JavaScriptCore: reduce difference with upstream…
<nekobot>
• pulkomandy (a5066e65): Remove unnecessary differences with upstream
<nekobot>
• pulkomandy (96cae3e9): SharedMemory.h: remove unneeded difference with upstream
HaikuUser has joined #haiku
HaikuUser has quit []
* OscarL-32
re-enables 19 Python tests that used to hang
<OscarL-32>
most still fail, but... progress? :-)
OscarL-32 has quit [Quit: Vision[]: i've been blurred!]
OscarL has joined #haiku
Aedil has quit [Ping timeout: 480 seconds]
<dovsienko>
I wonder why they didn't check errno
Nasina has quit [Read error: Connection reset by peer]
<OscarL>
that test in particular is more about how different locales can affec said string. the sillyness is assuming anything about such string in the first place.
Nasina has joined #haiku
diver has quit [Quit: Leaving.]
B2IA has quit [Quit: Vision[]: i've been blurred!]
B2IA has joined #haiku
KapiX has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
OscarL has quit [Remote host closed the connection]
janking has joined #haiku
<janking>
Wy can i not upload in forum, screenshot ?
Nasina has quit [Read error: Connection reset by peer]
Nasina has joined #haiku
Yoke has quit [Ping timeout: 480 seconds]
talos has joined #haiku
OscarL has joined #haiku
<OscarL>
"waiting for build package python3.12-3.12.10-1 to be deactivated" in what seems an infinite loop is not nice :-(
<brickviking[m]>
Incidentally, there's something I don't understand. If I run "hp <some-package-name>", it presumably compiles that package. Does it install it too, so it's visible by pkgman? Or is it merely plonked into a directory ready for installing by the Tracker?
<OscarL>
I'm starting to suspect that test_readline and test_pty are deadlocking each other.
<OscarL>
brickviking[m]: the resulting packages that hp creates are placed under the "packages" directory (unless you have specified a different path in haikuporter.conf).
<OscarL>
they are not auto-installed. if you want to do that after the build is complete... you can just "pkgman install <path to the newly created .hpkg file>"
<brickviking[m]>
ah, in other words: <where-source-is>/../packages?
<brickviking[m]>
(and yes, the .. was deliberate)
<brickviking[m]>
s/source/recipe/
<OscarL>
"base dir" is where you cloned the haikuports repo... say.... /Data/src/haikuports/
<OscarL>
then by default the packages end up in /Data/src/haikuports/packages
<brickviking[m]>
Ah huh. That somewhat correlates. I'm from a Linux background, so my stuff's in /boot/home/src/c/github/haikuports/haikuports
<OscarL>
my "haikuports.conf" file read: "PACKAGES_PATH=/Data1/haikuports/.packages", for example.
<brickviking[m]>
My Haiku is inside a VM, so I didn't tinker much. It's also on a machine that I'm meant to only be a guest on, so I don't press the issue.
<brickviking[m]>
I'm just glad I gave the VM 10GB to begin with.
<OscarL>
brickviking[m]: heck... many of of my installs are just 2 GB :-D
<brickviking[m]>
Ha. I couldn't get away with something that small. I've got about 4GB left of that 10.
<brickviking[m]>
About the only reason I can see for 2GB would be for single-package compilations
<OscarL>
before moving into an SSD, with a better (unified 32/64 bits) haikuports setup... I used a couple of bigger partitions.
<OscarL>
now I have my "main" one (beta5 64 bits, about 20 GB), and a bunch of smalller ones for 32 bits or for nightlies/custom haiku.hpkg testing, and a larger "/Data" partition where I centralize the cloned repos and such.
<brickviking[m]>
How many of those are bare-metal?
<OscarL>
all are dual purpose.... I can boot them either as bare-metal or from either VirtualBox or VMware Player 16.
<brickviking[m]>
Windows user?
<OscarL>
Mainly, yes, still have a couple of Linux installs around.
<brickviking[m]>
I gave up on Windows once I found out Microsoft was going the "Let's exfiltrate information from user's machine without the ability for them to refuse."