junaid__ has quit [Remote host closed the connection]
junaid_ has quit [Remote host closed the connection]
<Esmil>
Did anyone actually run Jisheng's patches for the TH1520/LicheePi 4A? My first attempt with 6.4 + the patchset didn't work as well as i'd liked: http://sprunge.us/8AAJsf
geranim0 has joined #riscv
junaid_ has joined #riscv
<conchuod>
Esmil: I would have hoped the other guy who resent it & sent the clock driver would have!
<Esmil>
Yes, I mailed both, but no reply yet
aerkiaga has joined #riscv
junaid_ has quit [Ping timeout: 252 seconds]
Tenkawa has joined #riscv
junaid_ has joined #riscv
awita has joined #riscv
Tenkawa has quit [Quit: Was I really ever here?]
geranim0 has quit [Quit: Leaving]
geranim0 has joined #riscv
junaid_ has quit [Quit: leaving]
junaid_ has joined #riscv
elastic_dog has quit [Ping timeout: 264 seconds]
khem has quit [Server closed connection]
khem has joined #riscv
elastic_dog has joined #riscv
junaid_ has quit [Ping timeout: 250 seconds]
elastic_dog has quit [Ping timeout: 260 seconds]
elastic_dog has joined #riscv
<palmer>
do you guys have any sort of setup that boots on it? I just took mine out of the box yesterday, but that's as far as I got ;)
junaid_ has joined #riscv
meta-coder has quit [Quit: leaving]
Tenkawa has joined #riscv
meta-coder has joined #riscv
aerkiaga has quit [Remote host closed the connection]
aerkiaga has joined #riscv
aerkiaga2 has joined #riscv
aerkiaga has quit [Read error: Connection reset by peer]
motherfsck has quit [Ping timeout: 245 seconds]
<muurkha>
jrtc27: related to yesterday's discussion of "independent standards bodies", CORBA and SOAP were just as disastrous as ACPI (and possibly UEFI, we'll see): https://news.ycombinator.com/item?id=36551401
<muurkha>
"The Internet had no time for multi-company, political garbage like SOAP and CORBA."
<Tenkawa>
CORBA was a nightmare for me in the 90's
<mps>
and XML
<Tenkawa>
I used it extensively in the telecom industry
<muurkha>
whatever the failings of XML, we can't blame them on XML being multi-company political garbage
<muurkha>
XML was three dudes
<mps>
I still have to use SOAP on $day_job but I rewrote all to plain XML
<muurkha>
hmm, I shouldn't erase Eve Maler from the history of XML, I guess, and apparently some guy named François Yergeau also has his name on the spec
<muurkha>
so five dudes and dudettes
<muurkha>
I think CORBA was almost universally acknowledged to be a nightmare
<Tenkawa>
muurkha: indeed!!!
<Tenkawa>
I know my entire group hated it
SpaceCoaster has quit [Quit: Bye]
SpaceCoaster has joined #riscv
<muurkha>
I think that's pretty much par for the course for software designs from "independent standards bodies"
<muurkha>
independent standards bodies are pretty okay when they're standardizing a design that is already known to work
<muurkha>
reconciling the appalling hacks in Cray C with the appalling hacks in Burroughs Large Systems C, things like that
bjoto has quit [Ping timeout: 240 seconds]
bjoto has joined #riscv
<muurkha>
or, relevantly here, clarifying the ambiguities in Waterman's dissertation and making enough minor political concessions that dozens of companies declare that they're going to support "RISC-V" instead of making up their own incompatible instruction set
<muurkha>
but structurally, an independent standards body originating a reasonable spec instead of an interoperability-nightmare monument to incompetence like ACPI or CORBA? it can happen but the chances are slim
jacklsw has joined #riscv
<dh`>
what you want isn't a standards body but a steering committee, or something like that
<dh`>
which is not necessarily a big difference, but the words you use to describe things matter
<dh`>
independent standards bodies also got us C++
loki_val has quit []
crabbedhaloablut has joined #riscv
<muurkha>
yeah, C++ is clearly in some sense workable, but surely you would agree it's also an interoperability nightmare
<dh`>
it has always teetered on the far boundary of workable
<dh`>
or something like that
<muurkha>
and we can expect whatever replaces it to be much simpler
<muurkha>
during the lifetime of the C++ standards committee I think we can sort of argue that it's been largely replaced by C
<muurkha>
well, and by Python, JS, etc.
<muurkha>
Java
motherfsck has joined #riscv
aerkiaga2 has quit [Remote host closed the connection]
junaid_ has quit [Remote host closed the connection]
junaid_ has joined #riscv
Leopold has joined #riscv
crabbedhaloablut has quit []
crabbedhaloablut has joined #riscv
<sorear>
I suspect it's less of a 501(c)6 problem and more of a "trying to be all things to all people" / "uncontrolled scope creep" problem
<sorear>
I don't think I've ever seen a large project go C++ -> C, and the reverse has happened a couple times
jacklsw has quit [Quit: Back to the real life]
<sevan>
is the ability to report CPU speed on Linux SoC specific or this functionality is currently missing? (tbh I'm actually on a release kernel on Ubuntu rather than stock upstream) that lack of reporting speed on the Allwinner D1 causing me problems
<sevan>
Tenkawa: thanks, that node doesn't exist for me
<Tenkawa>
it will likely be a different name per device
<Tenkawa>
first thing I tell people to do is try a find /sys | grep cpufreq and see if you have one
<Tenkawa>
you need both device tree and kernel entries to back it too
<sorear>
wouldn't cpufreq only be present if your board supports linux-controlled dynamic frequency scaling?
<Tenkawa>
sorear: not sure what its "called" but I just showed you a Starfive JH7110 running 6.3
zkrx has quit [Ping timeout: 240 seconds]
<Tenkawa>
that one was running 5.15 actually.. my other one runs 6.3
<sorear>
right, I'm looking at the d1 & nezha dts and there's nothing cpu clock related in there
<Tenkawa>
They dont have "any" frequency scaling?
<Tenkawa>
Ahh cpu
<Tenkawa>
er old cpu
<Tenkawa>
one of the first Risc-V's
BootLayer has quit [Quit: Leaving]
peepsalot has joined #riscv
<Tenkawa>
cheers all.. time to go fix a machine.... This should be.... interesting.
Tenkawa has quit [Quit: Was I really ever here?]
<muurkha>
sorear: yeah, I've never seen a large project go C++ -> C, and I have seen the reverse happen
awita has quit [Ping timeout: 260 seconds]
zkrx has joined #riscv
<muurkha>
but it's quite common for a large C++ project to be replaced by a somewhat smaller project in C, Java, Python, Golang, JS, etc.
<muurkha>
Rust!
<muurkha>
Microsoft Windows getting replaced by Linux is probably the best example of this
<muurkha>
also Symbian, Windows CE, and J2ME have been largely replaced by Linux, though iOS has made inroads at the high end
<muurkha>
and lots of places where the JVM has been replaced by Python or PHP, whose interpreters are written in C
<muurkha>
but no, you can't take a C++ project and port it to C, or to Rust or Python or JS, and it's usually pretty easy to port a C project to C++
<sorear>
the NT kernel is/was C
<sorear>
XNU is C++ though
<muurkha>
a lot of times it's more a piecemeal thing. I mean any Java program has C++ in it, right? because that's what HotSpot is written in! and commonly what happens is that an increasing fraction of new development is done in other languages, but without replacing the C++ parts of the project
<muurkha>
I didn't know that about the NT kernel! I thought it would have been C++, but I've never looked at it
<jrtc27>
who's replacing NT with Linux?
<jrtc27>
Microsoft sure aren't
<jrtc27>
people might take their projects to a totally different OS, but Windows is still NT
<sorear>
i personally beleive linux would be more maintainable if they used a properly defined c++ subset instead of YOLOing vtables and raii in C
<sorear>
this would be a harder sell a few years ago back when gcc could be built with a C compiler
<jrtc27>
I imagine they'd want a C++ that doesn't force you to cast from void *
<jrtc27>
because that drives me nuts
<jrtc27>
struct S *p = malloc(sizeof(struct S)); not compiling is just silly
<dh`>
we do need a new C, and it doesn't seem that rust is it
<sorear>
absolutely, because rust is the new C++
<muurkha>
jrtc27: most of the internet that was using NT replaced it with Linux
<jrtc27>
most consumers haven't
<muurkha>
most consumers have
<muurkha>
the number of Android users is larger than the number of Windows users ever was
<sorear>
i want someone to figure out what the C equivalent for Rust is, unfortunately there's a bit too much crosstalk between the lifetime inference and the rest of the type system to make a clean separation
<jrtc27>
they didn't replace a windows phone with a linux phone
<jrtc27>
they just bought a linux phone
<jrtc27>
most consumer PCs remain windows
<jrtc27>
hell most corporate PCs remain windows
<muurkha>
they replaced their consumer PC with a Linux phone, or never had a consumer PC
<muurkha>
it's true that most corporate PCs [that aren't phones] remain Microsoft Windows
<muurkha>
sorear: there's been a bunch of interesting academic work recently on minimalist Rust follow-ons like Koka
<sorear>
android phone but it's SBBR compliant
<dh`>
I keep wanting to do something about this (new C) but I don't know where the time's going to come from
<muurkha>
Golang is another source of inspiration, and it's what ken did when he decided it was time to build a new C
<dh`>
yeah and it's fine for some things but with a GC you can't use it for a kernel
<muurkha>
yeah, agreed
<muurkha>
and in the 30 years since C or C++ was the language all serious software was written in, pretty much everything that can tolerate a GC has abandoned C and C++
<dh`>
that's not quite true, but it's close
<muurkha>
so the Gophers are mostly ex-Python hackers or ex-Java hackers
<muurkha>
rather than "fat [gender-neutral informal term]s who know C++"
<muurkha>
(obligatory PC note: when I showed this to my friend Andrea she correctly protested that it was erroneous to exclude her from the guild on the basis of gender)
<muurkha>
anyway Koka, Zig, Nim, etc., are shaping up to be interesting
<sevan>
When you say Gophers, are you referring to people who work on the language, or just programmers in that language, in general?
<muurkha>
and if you can tolerate C++ you can probably tolerate Rust more easily, but it definitely doesn't appeal to the C crowd
<muurkha>
sevan: I meant people who are enthusiastic about programming in Golang
<muurkha>
so sort of in between
<dh`>
the problem I have with rust is the apparently toxic ecosystem it's spawned
<muurkha>
heh, yes
<muurkha>
lots of high-school clique drama
<muurkha>
people whose fundamental approach to life is based on hate and fear
<dh`>
also it's one in a long line of "you should use this because (evidence we don't understand what's wrong with either C or C++)"
<sevan>
There are various talks about the history of Go / decissions by Rob Pike. In the early days, one of the motivators was to get away from C++ because it was such an uphill battle for things that should be easy, I think the example he gave was fighting some template things. Organisationally, the drive was to have a language which they could give to the kids fresh out of college but they found it took several
<mps>
to rephrase "those who don't understand C are doomed to reinvent it, poorly" ;-
<sevan>
years to undo the damage from academia and for them to become safe in programming C++
<muurkha>
I think Graydon is probably pretty clear about what *he* didn't like about C++
<mps>
s/;-/;-)/
<muurkha>
but plausibly the current crew is a little less clear
<muurkha>
and Graydon has explained at some length how he would have taken a drastically different approach to evolving Rust
<muurkha>
can we get an F in the chat for Don Lancaster?
<sevan>
"Lancaster was an early advocate and developer of what is now known as print-on-demand technology. Lancaster produced his self-published books by re-purposing the game port of an Apple II to transfer PostScript code directly to a laser printer, rather than using a Macintosh running PageMaker. This enabled continuous book production using an inexpensive Apple II, rather than tying up an expensive Macintosh
<sevan>
until the print run was complete."
<muurkha>
(context: Macs didn't have multitasking and therefore a print spooler)
<sevan>
how about those Apple IIs? :)
<muurkha>
I think in a sense his most important invention was the Apple I
<sorear>
what, no print spooler as a desk accessory?
<muurkha>
Lancaster went a bit further than using his Apple ][ as a print spooler
<muurkha>
he just wrote his markup in PostScript
<mps>
isn't postcript some kind of forth language
<muurkha>
more like Lisp, but yeah
<dh`>
yes
<dh`>
no, it's a lot more postscriptish than lispy
<dh`>
er
<dh`>
a lot more forthish
<mps>
I liked forth long ago, even wrote one for Z80 embeded board
motherfsck has quit [Read error: Connection reset by peer]
<muurkha>
I don't think so. Forth is a statically-scoped, untyped language whose only dynamic memory allocation is a stack. PostScript is a dynamically-typed language with garbage collection, anonymous functions, dynamic scoping, and dictionaries
<dh`>
it is nonetheless based on stack ops like forth
<muurkha>
and interned symbols
<muurkha>
all they have in common is that they both use RPN syntax instead of conventional infix syntax
<mps>
dictionaries are also in forth
<muurkha>
Forth only has one dictionary
<muurkha>
in PostScript they're dynamically created, like dicts in Python or hashtables in Common Lisp
<mps>
well, one dictionary in time, but could have more of them
<muurkha>
(of course you can *implement* a hash table in Forth. you can implement anything in Forth)
<dh`>
idk, I think the stack-oriented semantics are the important bit
<muurkha>
that's just syntax
<dh`>
not really
<muurkha>
I guess another thing they have in common is that in both cases you can extend the syntax, as you can in Common Lisp
<muurkha>
by writing code that reads from the input stream
<dh`>
anyway, postscript is usually described as being from the forth family
<muurkha>
not usually, but occasionally, and in my view erroneously
<dh`>
usually IME
<muurkha>
from my point of view it's the same sort of mistake as thinking that Dylan is from the C family
<mps>
to me it also looks more forth than lisp, but I'm bad with lisp
<muurkha>
with respect to syntax extension, you have significantly less freedom to do this in PostScript; you have to do your metaprogramming old-Lisp-style at runtime by slinging around anonymous functions, not Forth-style (or Common-Lisp-style) by running code at compile time
<muurkha>
you can usually translate a PostScript program directly into Common Lisp with only surface changes, or vice versa. that is very much not true of Forth
<mps>
I can't comapare this, because I don't understand lisp much
<mps>
I trust your conclusion because you know both
<dh`>
I don't know enough about postscript
<dh`>
have tangled with it some, but mostly not in anger
<muurkha>
I'm no Lisp or Forth expert, but I have written a Scheme compiler in Scheme and a couple of crippled Forths, and although I've never implemented PostScript, I've written something you can charitably describe as a parametric CAD system in it
<muurkha>
PostScript used to be my favorite way to make simple graphics before I got hold of SVG
<dh`>
the reason an explicit value stack is not just syntax is that it exhibits behaviors you don't get with function application
<muurkha>
mostly bugs, but yes
<dh`>
like you can push stuff on the value stack and forget about them
<dh`>
s/them/it/
<dh`>
which is a bug yeah
<muurkha>
you can build up a list on the operand stack though
<muurkha>
the PostScript operators [ and ] (and << and >>) exploit this
<dh`>
also it seems like it would be hard to typecheck calls, but since the value stack is unique maybe session types would wor
<dh`>
k
<muurkha>
yeah, there have been some statically-typed languages using RPN syntax: Joy, Cat, and Kitten
<dh`>
... yeah, that should work fine, n/m
<muurkha>
but in general in PostScript you can [, run a loop to generate items, and then ] to package them up into a heap-allocated array
<muurkha>
you can't do quite that in Forth because you don't have a heap
<muurkha>
that is harder to statically type check
<dh`>
not really as long as they're homogeneous (same as lists anywhere else)
<muurkha>
I mean, your stack types have to take the variable length into account somehow
<muurkha>
you can do similar things in Forth if you add a heap (there's an optional vocabulary for this in dpANS FORTH) or if you're producing some values and then consuming them
<muurkha>
but it's not idiomatic the way it is in PostScript
<muurkha>
in Lisp you build up the loop directly on the heap
<muurkha>
*the list
<muurkha>
but from my point of view garbage collection and dynamic typing are two enormous differences that put them in different language families
<dh`>
yeah, but you can do that (have the stack type include a state "accumulating values of type T")
<muurkha>
Forth is in the same family as 386 assembly language, and PostScript is in the same family as Python
<mps>
muurkha: agree
motherfsck has joined #riscv
<mps>
afaik forth was intended for simple and real time apps, not for big complicated software
<muurkha>
or, attempting to be less offtopic, Forth is in the same family as RISC-V assembly language
<dh`>
when people talk about language families it's usually about some notion of style
<dh`>
and the value stack business is a very strong stylistic imprint
<mps>
actually there is RTX2000 CPU where forth is machine language
<dh`>
go and C and pascal and modula-* are all genearlly considered to be in the algol family even though they differ widely on a lot of things
<muurkha>
mps: Forth was intended for all apps, but the Forth belief system is most advantageous for simple and real-time apps
<muurkha>
I feel like the differences between C, Pascal, and MODULA-[23] are pretty minor
<muurkha>
except GC!
<dh`>
yeah
<muurkha>
it took me decades to appreciate Pascal's var parameters and procedure parameters, though
<dh`>
are there dynamically-typed languages commonly included in the algol family?