<SiFuh>
zorz: But you have to be a mass murderer to get the Nobel Peace Prize
<SiFuh>
Bill Gates openly told ABC’s “The View” that if USAID continues to be investigated and dismantled, a new pandemic could kill millions of people.
zorz has quit [Quit: leaving]
serpente has quit [Remote host closed the connection]
<remiliascarlet>
ukky: Intel is the most OpenBSD friendly. AMD works, but on mine I have problems with sndio. As for Nvidia, not even a chance.
<remiliascarlet>
Not sure about Intel Arc GPU's, but never had issues with Intel HD Graphics.
<SiFuh>
remiliascarlet: 3D?
<remiliascarlet>
SiFuh: I meant whether they work at all or not.
<remiliascarlet>
I have been able to use Blender on OpenBSD with an Intel HD Graphics chip, but Blender itself is very unstable on OpenBSD.
<remiliascarlet>
So better to run Blender on Linux.
<SiFuh>
Blender works fine here on AMD with AMDGPU
<remiliascarlet>
Besides, the latest version of Blender in the ports tree is extremely outdated, so it's clear that OpenBSD developers don't have any 3D modellers among them.
<farkuhar>
zorz: When Trump made the suggestion that he deserved the Nobel Peace Prize, did anyone in the audience press him about his promise to end the war in Ukraine on his first day in office?
<farkuhar>
remiliascarlet: The dictionary entry for "digest" points to the Latin root word "gerere" meaning "carry, bear", plus the prefix "dis". Now if Latin speakers pronounced the root word with a hard G, then we can blame the lazy speakers of Middle English for relaxing into the soft G pronunciation that you find so unintuitive.
<zorz>
farkuhar: Trump does business, Gaza Resort!
<zorz>
Trump does not care how to rule USA, only business.
<zorz>
actually Trump did something good, stocks of pfizer and moderna are in free fall. the cut of money aid to usaid hits soros. soros motherfucker fucked balkans.
<farkuhar>
Orthography changes slower than pronunciation, as a word migrates from its original language and gets picked up by speakers of other languages. remiliascarlet cited the example of digest (now pronounced with a soft G), while SiFuh cited the example of forecastle (whose middle syllables were dropped from the pronunciation, making it sound like fo'c'sle when spoken).
<farkuhar>
Another oft-cited example is featherstonehaugh, which admits the pronunciation "farnshaw", if I recall correctly.
<farkuhar>
SiFuh's latest pet peeve is with the overuse of acronyms, which he associates with "Fucking beerman language". But beerman's mother tongue is German, right? The more common criticism of German is that it forms compound words too readily, not that it overuses acronyms.
<farkuhar>
I'm not a big fan of German's choice to keep the word "Bleistift" for pencil, which prominently triggers the association with a long-discontinued practice: using lead (Blei) for the marking element. Surely by now they could have adopted a word without such anachronistic associations.
<ppetrov^>
somehow I have trouble with tar.xz packages, it expects tar.gz
serpente has joined #crux-social
<ppetrov^>
ok, i figured it out...
<ppetrov^>
i had space before PKGMK_COMPRESSION_MODE="xz" in /etc/pkgmk.conf
<farkuhar>
ppetrov^: No update that I'm aware of. But you raise a good point about needing a better parsing algorithm, for programs that refer to pkgmk.conf yet are not themselves written in bash.
<farkuhar>
As darfo suggested, maybe the safest solution is to spawn a bash shell, which will guarantee that the program is on the same page as pkgmk.
<ppetrov^>
is it possible to speed up ports -u?
<ppetrov^>
my r4-modules repo has over 500 ports and it takes quite some time
SiFuh-KL has joined #crux-social
<SiFuh-KL>
farkuhar: SiFuh is here
<ppetrov^>
farkuhar, SiFuh-KL is here
<SiFuh-KL>
Hahahaha
<ppetrov^>
hahahahaha
<SiFuh-KL>
Gloucester is easy to pronounce if you are Scottish
<SiFuh-KL>
farkuhar: gloss-te-sheer
<SiFuh-KL>
ppetrov^ no need to prove your hahas are longer than mine
<ppetrov^>
ha
<SiFuh-KL>
I'm in KL because an Audi owner of a Q6 needs a mechanic. No one here wants to touch his piece of shit car. So I am here to inspect and possibly repair it
<SiFuh-KL>
Looks like plenty of free beer coming soon
<SiFuh-KL>
My truck is 35 years old and still going. Damn Toyota is good
<SiFuh-KL>
I think you said his car is 4 years old already falling apart. And hopefully if I managed to fix it he will sell it to some unsuspecting gullible driver
<SiFuh-KL>
You/he
<SiFuh-KL>
farkuhar after you abandoned me in Penang I went to a side street bar. Today in the pub they were trying to remember the name of it fortunately I knew the name and was able to interject
<SiFuh-KL>
farkuhar: SiFuh will quit
SiFuh-KL has quit [Quit: SiFuh-KL]
<farkuhar>
ppetrov^: It should be possible to speed up ports -u. The shell script currently doesn't attempt to divide up the work into multiple parallel threads, so that's an easy place to start looking for possible speedups.
<farkuhar>
On the Wiki page PkgutilsIdeas there's a reference to jw's early tests comparing the performance of 'ports -d' and 'prt-get diff', or 'ports -l' and 'prt-get listinst'. In both cases the shell script was slower by a noticeable margin.
<farkuhar>
Hence the recommendation to get rid of the listing and diff functionality of the ports script, and keep it around only for the repo sync operations. But if you get rid of 2/3 of the functionality, why not drop the script entirely, and let prt-get handle even the sync operations?
<ppetrov^>
thanks farkuhar
<farkuhar>
ppetrov^: Here's another pkgmk.conf that is too complicated for the naive parsing of pkg-get, and a possible reason to adopt darfo's strategy "just spawn another bash instance". https://0x0.st/8P-A.conf
<farkuhar>
Defining PKGMK_COMPRESSION_MODE in this way will also break prt-get's parsing strategy. None of the high-level tools tries to anticipate the possibility that PKGMK_COMPRESSION_MODE could vary from port to port, depending on something so unessential like the maintainer field.
<farkuhar>
As soon as jw realized that his new creation prt-get would be consulting the same config file as a tool that operates at the individual port level, he could have insisted that the common data fields (compression mode, package directory) be isolated in their own config file, untouchable by the playfulness and creativity that might be permitted in a file only seen by pkgmk.
<farkuhar>
An extra layer of insulation around the config fields that affect tools at multiple levels (individual port level versus entire repo level) would have prevented the need to spawn an extra bash shell (the most reliable way to ensure that the higher-level tools are on the same page as pkgmk).
<farkuhar>
I'm struggling to come up with a way to fix pkg-get's parsing algorithm so that it can make sense of the pkgmk.conf posted above. "Just spawn another bash shell", okay, but on many deployments just *one* more shell would be insufficient. You'd have to ensure that you capture the full range of possible outputs from the shell function compression_per_maintainer().
<ukky>
remiliascarlet: thanks for feedback. I doubt that I will run OpenBSD on any of my systems, but it is good to know which HW gets more coverage among all BSD.
<farkuhar>
`grep getPkgmkSetting *.cpp` in the prt-get src directory reveals that jw only extracted two lines from pkgmk.conf: the lines defining PKGMK_PACKAGE_DIR and PKGMK_COMPRESSION_MODE. If either of these variables has a hidden dependence on Pkgfile variables like $name, then jw's code is not up to the task.
<farkuhar>
SiFuh: Wait, wouldn't gloss-te-sheer be closer to the pronunciation of Gloucestershire (not Gloucester as you wrote above)?