<sen>
SArpnt[m]: I've been surprised how daily-driver it's become for me, it's simple in a pleasant way, never advertises at me, and never makes me login using The Cloud or anything like that.
HaikuUser has joined #haiku
HaikuUser has quit []
<waddlesplash>
personally I find it very fun that when people ask me "so how many people use Haiku on a daily basis?" I can tell them "I have no idea, we don't collect that data"
<waddlesplash>
the most I can tell you is website analytics (but these respect DNT so who knows how accurate they are) and package download hit counts (but eventually these will go through a CDN/mirror network and then we probably won't know that either)
<waddlesplash>
and at the moment SoftwareUpdater doesn't run automatically, either, so it's entirely possible to use Haiku on a daily basis but contact our servers practically never
<SArpnt[m]>
i mean i don't plan on using it as a daily driver when i already have arch linux set up
tuaris has quit [Remote host closed the connection]
AlaskanEmily3 has quit [Remote host closed the connection]
zayd has joined #haiku
MisthaLu has quit [Read error: Connection reset by peer]
HaikuUser has joined #haiku
HaikuUser has quit []
<nipos>
waddlesplash: Just tried booting with the latest Nightly hrev58209 on the 128GB RAM server and it doesn't work.But I think this hrev may still be behind the one were you fixed it.Will try again tomorrow or in a few days.
AlaskanEmily3 has joined #haiku
AlaskanEmily3 has quit [Remote host closed the connection]
OscarL has joined #haiku
diver has joined #haiku
AlaskanEmily3 has joined #haiku
* OscarL
also waits for hrev58221 to be available to see if it helps with his cursed Atom N2600 headless netbook.
<OscarL>
morning Begasus.
diver1 has quit [Ping timeout: 480 seconds]
<Begasus>
morning OscarL :)
<Begasus>
no real progress on the downloads for nightlies
<OscarL>
Playing with "pyinstrument" and haikuporter. Shows that the biggest timesinks are the scriptlet/bash calls, so not much point in trying to optimize the python code :-/
<Begasus>
no slacking! :)
<OscarL>
ran "pyinstrument --use-timing-thread haikuporter --lint" two consecutive times. First took 570 seconds, ~480 secs of that are the bash calls. ~80 secs is reading from the "repository/" cache.
<Begasus>
maybe it's optimized enough already?
<OscarL>
second time it took 433 secs. with almost 400 secs for the bash calls, and about 25s reading from "repository" (already cached by the OS).
<OscarL>
Begasus: that's what I was trying to figure out. I had some changes that use picked objects instead of plain files under repository... runs about the same speed when caches are "cold", but on subsequent runs, the OS I/O cache makes the "plain files" version a tad faster :-D
<OscarL>
wanted to run "proper" benchmakrs... like in "pyperformance (The Python Benchmark Suite)"... but that needs pyperf, and that needs psutil. Need to see how much of the latter is needed for pyperf to work reliably.
<OscarL>
nipos: thanks for your contributions :-)
<nipos>
No problem,always happy to help :)
<Begasus[m]>
+1 (although I only use black screen for screensaver) :)
* OscarL
changes Begasus[m] screensaver to Icons while he is distracted.
vdamewood_ has joined #haiku
diver has quit [Read error: Connection reset by peer]
janking has joined #haiku
<Begasus>
nogo :P
<Begasus>
I've got a special block OscarL script now :P
diver has joined #haiku
OscarL is now known as BiPolar
<Begasus>
with an alias ... :P
<BiPolar>
"nickname is registered and protected" :-(
BiPolar is now known as OscarL
<OscarL>
Begasus: maybe I should try something more subtle, like: /nick perfectly_safe_zlib_install_now
diver has quit [Read error: Connection reset by peer]
<Begasus>
heh
vdamewood has quit [Ping timeout: 480 seconds]
<Begasus>
add a note to the recipe: "handle with care!" ;)
diver has joined #haiku
<Begasus>
grabbing catdoc-0.95-1-x86_64.hpkg and moving it to /Share/haikuports/packages/catdoc-0.95-1-x86_64.hpkg
<Begasus>
finetuning ...
arjen_ has joined #haiku
diver has quit [Read error: Connection reset by peer]
<nekobot>
[haiku/haiku] 091140b15a92 - Update translations from Pootle
diver has quit [Read error: Connection reset by peer]
diver has joined #haiku
<OscarL>
Begasus: "pyinstrument" can output results in .html... I was wondering why Web+ was so slow... the .html file was 25+ MB :-D
<Begasus[m]>
lol
<Begasus[m]>
should use sphinx for that :)
<OscarL>
granted... giving it "hp --lint" to profile is a bit on the extreame side (given the number of recipes we have), so no wonder the results are so large :-D
<OscarL>
but it handles it like a champ (given that you have at least 2 GB of RAM), so... well done pyinstrument!
<OscarL>
welp, even with pyinstrument... can't really tell appart any meaningfull performance diff bewteen upstream haikuporter, and the version with my changes. Both about the same after a reboot, and after I/O cache warm up.
<OscarL>
I do like having less files on disk, so... I'll keep using my version, I guess :-)
<Begasus>
having a diversed haikuporter doesn't break potential builds on the buildmasters?
<OscarL>
I'm just using the same data, but instead of writting it as .DependencyInfo files... I just store them in a dictionary that gets saved to file and read when needed.
<botifico>
[haikuports/haikuports] Begasus 819bc8e - catdoc, new recipe (#11208)
<OscarL>
changes are pretty isolated, and it is just on the "cache" side of things, if I messed up, worst case, I make my builds slower :-)
<OscarL>
Tracker's PgUp/PgDn is acting really weird on this "haikuports/repository" dir with 6270 files :-/
<OscarL>
in listview mode, being at the "top of the list"... first two PgDn presses seem to work normally, but the third sends me ritht to the last file.
<OscarL>
and then PgUp only "works" 3 times, then stops doing anything :-D
<Begasus>
who ever scrolls in that directory? :P
<OscarL>
Where's Skipp_OSX when you have a bug for him? :-P
<OscarL>
Begasus: /me, of course!
<Begasus>
figures :)
<OscarL>
mmm, no wonder the diff was small. The changes I was actually testing only included the using a picled dict for to replace the ".skipped/*" files :-)
<OscarL>
wonder where I left the rest of my commits, lol :-)
<Begasus>
rofl
<Begasus>
in the clouds? )
<OscarL>
... of my dreams, maybe? :-D
<Begasus>
heh
nephele has joined #haiku
<nephele>
hi
<OscarL>
hello nephele.
HaikuUser has joined #haiku
HaikuUser has quit []
<nephele>
zardshard: heyhey. I think I got most stuff finished for the keyboard yesterday. Only remaining problem was that the WebViewBase would not want to accept the focus, so i could not test it yet :)
<Begasus>
*** Failed to find a match for "attica6_devel-6.7.0~rc1-1-x86_64.hpkg": Name not found
<OscarL>
Begasus: pkgman has some odd edge cases, like "pkgman update haiku*" wanting to *install* package I don't have, instead of just updating the haiku*whatever packages that are already installed :-)
<Begasus>
that's run from the path as shown OscarL
<nephele>
don't use wildcards then :P
<Begasus>
nephele, last paste was without :P
<nephele>
I ment oscarl ;)
<OscarL>
:-P
<Begasus>
ah sorry :)
<jmairboeck>
Begasus: something like this should work: echo *_devel-6.7.0~rc1-1-x86_64.hpkg | cut -d- -f1 | xargs pkgman uninstall -y
<nephele>
yeah. You better apologize! /s
<Begasus>
just did :P
<Begasus>
checking jmairboeck
<nephele>
Auch besser so! /s
<Begasus>
only takes one argument it seems there jmairboeck
<nephele>
WebViewBase crashes (the part rendering the WebView). I'm wondering if I can add an ilustration of a wrapped tuna and write "WebProcess is swimming with the fishies ..."
<Begasus>
Karma :P
<OscarL>
while it would be nice if pkgman uninstall accepted package filenames... why on earth would you try " pkgman uninstall *_devel-6.7.0~rc1-1-x86_64.hpkg" and not "pkgman uinstall attica6_devel" Begasus? :-D
<OscarL>
more like a rethorical question... no need to explain really :-)
<Begasus>
OscarL, the idea is to uninsttall any *6.7.0* devel package in there (67 in total) :)
<OscarL>
(I too have wanted to do things in a "this obviously should work, why doesn't it?" more than once)
<OscarL>
I remember being really pissed once at ObjectPascal... I couldn't understand why of "+" did string concatenation... "-" failed to substract characters from a string :-P
<Begasus>
like when having 20 *python310* packages you wanted to uninstall in one row
<nephele>
why do keep trying to use "pkgman i" as a shorthand for "pkgman install"
<OscarL>
if "foo" + "bar" == "foobar", then surely "foobar" - "bar" == "foo", right? Right? :-P
<jmairboeck>
nephele: "pkgman ins" still works, "pkgman i" broke when "info" was added
<nephele>
jmairboeck: oh! so this *was* supported at some point
<jmairboeck>
the prefix must be unique
<nephele>
Well, that makes me happy a bit. means i actually did use this
<nephele>
and not carried over from some other OS
<jmairboeck>
just "i" is now ambiguous
<nephele>
:)
<nephele>
maybe instead of shoving the help in ones face pkgman could say "i is ambigious between info and install" or something
<nephele>
that would make it immidiently obvious what has happened
<OscarL>
Begasus: I see. Welp, works for *_python3* because that matches the canonical package names (excluding version, arch, extension).
<OscarL>
Begasus: you might want to open a new enhancement ticket for uninstall to accept filenames too.
<OscarL>
Next time take a screenshot of a Tracker window of that folder, run said screnshot throug an OCRTranslator, cat resulting text file. Got it! :-P
<nephele>
OscarL: why don't you implement deleting packages overs to uninstall them first, and only deleting them as an alternative? :P
<nephele>
offers*
<OscarL>
heh, "shopt -s nullglob" / "for f in *" again on that link nephele shared. Universe trying to tell me to not forget about, it something :-)
<OscarL>
s/it/or/
<nephele>
Clearly, should implement that new shell for haiku and forget about all this
<nephele>
saw a "meme" picture that said programmers will be happy about learning hard things, and then resist attempts to make it easier, rings kind of true xD
<OscarL>
nephele: couple months ago I was lamenting having to have found that nullglob thing... and wondering where things took such a wrong turn for me :-P
<nephele>
OscarL: honestly shell script is so wierd. despite it beeing so incredibly difficult ressources are spread out all over
<nephele>
there is no "So, you think you know shell script? *laugh crying emoji*" book :)
<OscarL>
re: package uninstall instead of deletion... while I *might* be able to do it... with my poor skills and focus issues... better left it to smarter folks than me :-)
<nephele>
I discovered yesterday that the "streaming" context helped me focus a whole lot
<nephele>
i usually get distracted very easily
<SArpnt[m]>
i just use fish shell, it simplifies all the mechanics of a shell
<nephele>
the fish shell can't help there really, because the problems already begin with the commandline interfaces of POSIX programms
<nephele>
i.e how they interpret arguments passed to them
<nephele>
how they can't get arguments after start (except in specialized programms)
<nephele>
how there is no "streams"with ordered data except for "hook up stdin and stdout" (but just "bytes"
<SArpnt[m]>
i'm not sure what you would want either of those for
<SArpnt[m]>
changing arguments midway through doesnt even make sense, thats like asking to change command midway through
<SArpnt[m]>
or a cake recipie that says to remove the oil halfway through baking
<Begasus>
k, KF6 6.7.0 installed now :)
<nephele>
How so? Haiku programms do this all the time. ShowImage can receive new images to view after starting. But you can't "just" give a new website adress to firefox on linux without the programm implementing this with a second executable and a shared memory interface or something
<SArpnt[m]>
most shell programs do one thing then quit
<nephele>
I was talking about "commandline interfaces of POSIX programms"
<SArpnt[m]>
all the images in showimage are seperate
<Begasus>
well, ls helped out producing a clean list I could use nephele
<Begasus>
OscarL, might add a feature request for this
<SArpnt[m]>
what posix command do you want to change arguments midway through and how is it different from just stopping and running it again with the new arguments
<nephele>
No? It's just one process with severall windows. It got the "command" to open a second image *with ordered data* after running
<nephele>
dd for example?
<nephele>
Tell it to start caughing up progress data
<nephele>
since linux also doesn't do SIGINFO...
<SArpnt[m]>
caughing?
<OscarL>
Would be nice if "status=progress" would be dd's default if isatty().
<coolcoder613>
mm
<coolcoder613>
good idea
<coolcoder613>
you should patch it
<nephele>
OscarL: implement SIGINFO ;)
<SArpnt[m]>
dd has a pretty bad interface
<coolcoder613>
true
<coolcoder613>
but you get used to it
<nephele>
You shouldn't have to
<coolcoder613>
you shoudln't have to get used to vim, too...
<SArpnt[m]>
anyways i dont think launch arguments in general could be changed midway through thats just not how it works
<SArpnt[m]>
normally i would say just change how it works
<coolcoder613>
Maybe a gnu long options sort of change
<SArpnt[m]>
but they are launch arguments and i'm not sure what system you would replace it with
<SArpnt[m]>
also in most cases changing thwm midway through wouldnt work at all
<nephele>
Yes, that is the point. *fish* cannot ever fix this deficiency while remaining compatible with posix. But there is nothing stopping us making a seperate interface analogous to the usecases sh does, without using the POSIX interface
<SArpnt[m]>
dd could start showing progress later, but it couldnt change the file being written
<SArpnt[m]>
most programs use the arguments for things that couldnt change later
<SArpnt[m]>
and as long as its possible to have things that can't change later, programs will be able to have things that should be changable but arent
<nephele>
some do, some don't. it does not matter. The only case where changing "launch args" does not make sense is for one shot operations like creating a single dir, a single file, etc.
<nephele>
the replicant for "workspaces" from my stream yesterday is in the middle of my screen now xD
duncsauce has joined #haiku
duncsauce_ has quit [Ping timeout: 480 seconds]
duncsauce has quit [Remote host closed the connection]
duncsauce has joined #haiku
<sen>
different projects have different preferences for this so figured I’d ask before proceeding. does it make more sense to put in one bug for “multi-battery issues” and put all the details there (idle batteries show as “charging”, low battery warnings happen when each battery gets low even if there’s still fully charged ones, etc.) or one bug per actual issue?
<OscarL>
If they all have the "multi-battery setup" in common, using one ticket for all would be my move. It can be broken up later on, if needed.
<OscarL>
(spoken just as a noisy user, mind you)
<nephele>
usually we pick one bug per issue, but in this case i think one ticket is fine since they all have the route cause of not having multi battery in mind
<nephele>
and: in general: don't worry too much about it, if something could be organized better on our bugtracker we will triage this ourselves, so the ocasional mistake is totally fine :)
<sen>
sounds good, will combine, thanks :) will probably end up submitting patches to fix things too, but cataloguing what’s actually wrong is a start.
duncsauce_ has joined #haiku
duncsauce has quit [Ping timeout: 480 seconds]
<OscarL>
Begasus[m]: your "no_valid_boot_volume.jpg" shows several "size xxx.999 MB". Besides your boot problems, seems 0.001 MB keeps getting "stolen" on some conversion somewhere :-P
<Begasus>
no expert there OscarL, it's just done with Etcher, so no fancy stuff ? :(
<OscarL>
wait... is that the internal or the external drive?
<Begasus[m]>
testing ... new neochat ...
<OscarL>
(doesn't sounds like the external one... if it has a Microsoft reserved partition)
<Begasus>
I think the bottom part is from the thumbdrive OscarL, above is the internal drive
<OscarL>
ah, so, that's the info you get while trying to boot from pendrive, and it scans all those partitions, right?
zardshard has left #haiku [Error from remote client]
<Begasus>
right OscarL
zardshard has joined #haiku
<OscarL>
still not clear how those partitions are organized. [1] and [4] look like Win reserved / Win Recovery. [2] is the Windows install then I assume?
<Begasus[m]>
3 = BFS data partition
<Begasus[m]>
5 was the BFS partition (deleted/reformatted)
<OscarL>
then why it still shows up?
<OscarL>
also, then it is all on the internal drive, no?
<Begasus[m]>
5 isn't BFS anymore, it still shows the full list of partitions on this drive (internal)
<Begasus[m]>
it's the part after 5 that's related to the USB thumbdrive (imho)
<Begasus[m]>
"recognized but couldn't scan"
<OscarL>
I would wipe 5 with GParted or at least change the partition tipe to Linux, for example. Might be enough to not confuse the Haiku boot loader?
<Begasus[m]>
had it done with GParted earlier :)
<Begasus[m]>
but could have a go again (sometime later) :)
<OscarL>
the fact that it still has "name: Haiku" should be an indication that it is not really wiped.
<Begasus[m]>
that "just" a name :)
<Begasus[m]>
had it as "NO NAME" before
<Begasus[m]>
doggies ... :)
<OscarL>
Not seeing the ESP on the internal drive either :-/
<OscarL>
oh well... food time. Laterz!
OscarL has quit [Quit: Gone with the wind]
Begasus has quit [Quit: Vision[]: i've been blurred!]
Begasus has joined #haiku
nephele has quit [Quit: Vision[]: i've been blurred!]