Re: [9fans] Olimex: these guys are keen electronic engineers.
On 2/24/21, tlaro...@polynum.com wrote: > > FWIW, I bought Olimex-lime2 (ARM) (severals) and I'm more satisfied with > these than with > Raspberries (I installed NetBSD on this, plan9 was not tried). > > So if the RISC-V is on the same level of quality, it should be certainly > worth. Exactly my point, thank you for corroborating. I've been eyeing their DIY laptop, but for that price (and custom complications) I may prefer to invest in the 3D printer they list as in stock at quite an old price (I don't expect miracles, but it's worth a try). I think these guys operate in a rarefied atmosphere of custom design and they have a working formula. I wish I could contribute in a significant way to their success. I have a long wish list revolving around their products and philosophy, just not the funds. Lucio. -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/Tef717f57ede82d4f-Mf8206ac5ede4d151a6f9b6d0 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] Olimex: these guys are keen electronic engineers.
On Wed, Feb 24, 2021 at 07:50:59AM +0200, Lucio De Re wrote: > They say: > > > This is the first Espressif product with RISC-V core, the datasheet is on > > their web. > > > > This is also the first SOC with RISC-V core we have access to, so we are > > excited to learn > more the ISA on low level. > > > > Any resources to recommend? > > So far I have been shy to recommend Plan 9 to them, a little less shy > to recommend Olimex to 9fans (somewhat long ago, my memory may be > lying to me). > > In any case, for the likes of Richard Miller and other wizards, this > is the URL for this specific > posting: > > https://olimex.wordpress.com/2021/02/23/hello-risc-v-we-got-samples-of-the-new-esp32-c3-module-and-it-is-only-13x17-mm/ > > >From there, many recent and less recent developments can be > discovered. They have a neat catalogue and apply pretty good QC. For > those in the EU, they operate from Bulgaria. Their English can be > surprising. FWIW, I bought Olimex-lime2 (ARM) (severals) and I'm more satisfied with these than with Raspberries (I installed NetBSD on this, plan9 was not tried). So if the RISC-V is on the same level of quality, it should be certainly worth. -- Thierry Laronde http://www.kergis.com/ http://kertex.kergis.com/ http://www.sbfa.fr/ Key fingerprint = 0FF7 E906 FBAF FE95 FD89 250D 52B1 AE95 6006 F40C -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/Tef717f57ede82d4f-M79a9c710c32122473449f811 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
[9fans] Olimex: these guys are keen electronic engineers.
They say: > This is the first Espressif product with RISC-V core, the datasheet is on > their web. > > This is also the first SOC with RISC-V core we have access to, so we are > excited to learn > more the ISA on low level. > > Any resources to recommend? So far I have been shy to recommend Plan 9 to them, a little less shy to recommend Olimex to 9fans (somewhat long ago, my memory may be lying to me). In any case, for the likes of Richard Miller and other wizards, this is the URL for this specific posting: https://olimex.wordpress.com/2021/02/23/hello-risc-v-we-got-samples-of-the-new-esp32-c3-module-and-it-is-only-13x17-mm/ >From there, many recent and less recent developments can be discovered. They have a neat catalogue and apply pretty good QC. For those in the EU, they operate from Bulgaria. Their English can be surprising. Lucio. -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/Tef717f57ede82d4f-M7e82a6b6e681265d7722f6fe Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] sshnet and the go dns resolver
thanks all for the suggestions. digging through the source of sshnet.c and the go net library i think i understand. the plan9 runtime library assumes, and insists the string returned from /net/cs contains a valid ip address. the code in sshnet which serves a /net/cs does a local ndb lookup, and if that fails, returns the dns name it was given. i guess this is to allow name resolution to be tried at the remote host. i shall experiment a bit more. -Steve -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T774469ff9f943e45-M892fd19deed0c0a02dd942a8 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] APL
tlaronde pointed me at the APL that shipped in the contrib directory in 4.3BSD. In hindsight I suspect that was the version I spun up at Athabasca U way back when (1989ish). I was quite surprised to see that a substantial chunk of it managed to compile 'out of the box' on OpenBSD 6.8 (albeit with a flood of warnings :-)): : lyndon@broken:/u/lyndon/src/apl/4.3/usr/contrib/apl/src; ls *.[co] Llx.ca4.c a8.o ac.o ag.c ak.c ao.o ax.c gamma.c a0.c a5.c a9.c ad.c ag.o al.c aplcvt.c ax.o lex.c a1.c a6.c aa.c ae.c ah.c am.c aq.c ay.c tab.c a2.c a7.c aa.o ae.o ai.c an.c at.c az.c xed.c a3.c a7.o ab.c af.c aj.c an.o at.o az.o y.tab.c a3.o a8.c ac.c af.o aj.o ao.c aw.c cata.c y.tab.o Seems like a viable candidate to base the port on. --lyndon -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T233ff29f045d64a9-Mb2cc2b7e0f346fbbd83e53dd Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] sshnet and the go dns resolver
I’ve noticed similar problems with go programs and resolving a lookup. There are go programs that explicitly look for “/etc/resolv.conf” to grab the nameserver to use, though it would be nice if the developer were using the default net package for lookups. > On Feb 23, 2021, at 3:34 PM, Steve Simon wrote: > > hi, > > the senerio - i have a plan9 terminal, a mac laptop. the laptop connects to a > vpn. > > i write in go, and thus far i use sshfs to mount the mac’s filesystem and > edit my code. i build and run the code on the mac. > > i thought i could use git9 via sshnet to work natively on plan9 > > most of this is fine but when go tries to access mod files for packages on > the vpn i get a “no sutiable address found” error for proxy.golang.org. > > i can resce the address i need (over the vpn) using ndb/dnsquery however. > > if i try and build without my sshnet vpn connection then i get a different > error which refers to a repository on the vpn. > > so. i think the native resolver in the go libraries (and thus in the go > executable) do not play nicely with sshnet. > > my guess (only a guess) is go is trying to do its own name resolution using > udp port 53 and sshnet does not support udp. > > anyone any ideas? > can i convince go to use plan9’s own name resolution? > can ssh (and thus sshnet) forward udp or is that not possible? > > thanks for any ideas > > -Steve > -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T774469ff9f943e45-Mab4a9e43134cbeda9523e532 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] sshnet and the go dns resolver
The Go net package uses /net/cs and /net/dns. I see sshnet doesn't create /net/dns. That could be the issue. fhs On Tue, Feb 23, 2021 at 4:35 PM Steve Simon wrote: > hi, > > the senerio - i have a plan9 terminal, a mac laptop. the laptop connects > to a vpn. > > i write in go, and thus far i use sshfs to mount the mac’s filesystem and > edit my code. i build and run the code on the mac. > > i thought i could use git9 via sshnet to work natively on plan9 > > most of this is fine but when go tries to access mod files for packages on > the vpn i get a “no sutiable address found” error for proxy.golang.org. > > i can resce the address i need (over the vpn) using ndb/dnsquery however. > > if i try and build without my sshnet vpn connection then i get a different > error which refers to a repository on the vpn. > > so. i think the native resolver in the go libraries (and thus in the go > executable) do not play nicely with sshnet. > > my guess (only a guess) is go is trying to do its own name resolution > using udp port 53 and sshnet does not support udp. > > anyone any ideas? > can i convince go to use plan9’s own name resolution? > can ssh (and thus sshnet) forward udp or is that not possible? > > thanks for any ideas > > -Steve > -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T774469ff9f943e45-M6e96a817ce5e09891b086877 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
[9fans] sshnet and the go dns resolver
hi, the senerio - i have a plan9 terminal, a mac laptop. the laptop connects to a vpn. i write in go, and thus far i use sshfs to mount the mac’s filesystem and edit my code. i build and run the code on the mac. i thought i could use git9 via sshnet to work natively on plan9 most of this is fine but when go tries to access mod files for packages on the vpn i get a “no sutiable address found” error for proxy.golang.org. i can resce the address i need (over the vpn) using ndb/dnsquery however. if i try and build without my sshnet vpn connection then i get a different error which refers to a repository on the vpn. so. i think the native resolver in the go libraries (and thus in the go executable) do not play nicely with sshnet. my guess (only a guess) is go is trying to do its own name resolution using udp port 53 and sshnet does not support udp. anyone any ideas? can i convince go to use plan9’s own name resolution? can ssh (and thus sshnet) forward udp or is that not possible? thanks for any ideas -Steve -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T774469ff9f943e45-Me6b7b01e94fd355714ebf049 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] 2c/2l make sense, but why 1c/1l?
I don't believe a 68000 compiler was ever released by the labs but there may have been one - some blit terminals had 68000s (and maybe gnots?) so its plausable. There was a port of the plan9 compilers to the VAX but I think its sourcecode was lost (jmk found an executable some years). -Steve -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/Tf34475f1bb69674a-M93c386537b2c045b411b3708 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] APL
On Tue, Feb 23, 2021 at 02:57:44AM -0700, arn...@skeeve.com wrote: > This is getting off topic ... > Yes... ;-) > > > There was an interpreter for P-code and (I think later) a compiler > > > for the Vax. You'd have to port it to current architectures, and > > > compiling TeX would probably make TeX run more slowly than the C version. > > > > > > The Berkeley Pascals were some of the compilers used for "Software Tools > > > in Pascal". > > > > The Pascal version would probably be a bit slower. And it would be more > > an alternative to verify the code than a primary way, since in fact > > D.E.K. has not written the program in some Pascal but in Algol, a high > > level abstract description, the wizardry being in the data structures. > > It's Pascal, but in literate form with WEB. I've read "Tex: The Program". :-) What I wanted to say is that D.E.K. has put aside all that is "system dependent". Yes it's Pascal, but the very least common denominator with all that is system dependent isolated so that porting (or translating) is easy with the change files and the chunks identified. I never managed to like Pascal (the language taught when I was---vaguely---in college) and I, first, would have liked that it was CWEB and not WEB (C and not Pascal). But, indeed, it is an abstract way of describing and prevent a myriad of hasardous hacks. So D.E.K.'s was not a bad choice alltogether... -- Thierry Laronde http://www.kergis.com/ http://kertex.kergis.com/ http://www.sbfa.fr/ Key fingerprint = 0FF7 E906 FBAF FE95 FD89 250D 52B1 AE95 6006 F40C -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T476a1d7b83269775-Mef5e8af94c52c9c5c0bd15b3 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] APL
This is getting off topic ... > > There was an interpreter for P-code and (I think later) a compiler > > for the Vax. You'd have to port it to current architectures, and > > compiling TeX would probably make TeX run more slowly than the C version. > > > > The Berkeley Pascals were some of the compilers used for "Software Tools > > in Pascal". > > The Pascal version would probably be a bit slower. And it would be more > an alternative to verify the code than a primary way, since in fact > D.E.K. has not written the program in some Pascal but in Algol, a high > level abstract description, the wizardry being in the data structures. It's Pascal, but in literate form with WEB. I've read "Tex: The Program". :-) > And, indeed, only the control flows are being translated from pseudo > Pascal to C, the core---the data structures---being handled by ad-hoc > code. Could be, I'm not familiar with how web2c works. > And for the architectures, like other compilers, the aim would be to > convert to some intermediate language (perhaps assembly) and to borrow > the back-ends. I think the Pascal compiler used the PCC back end, but I no longer remember for sure. If so, you might could hook it up to the revived PCC project. Although it sounds like a fun project, there are probably better uses for your time. :-) Arnold -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T476a1d7b83269775-Mc80e8615b718f628a8468339 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] APL
On Tue, Feb 23, 2021 at 01:58:16AM -0700, arn...@skeeve.com wrote: > tlaro...@polynum.com wrote: > > > On Mon, Feb 22, 2021 at 09:44:54PM +, Charles Forsyth wrote: > > > I'm fairly sure Thompson wrote it on sabbatical in Berkeley. I think he > > > also wrote the first version of a Pascal compiler. > > > Pascal isn't a difficult language but I remember that compiler having an > > > unusual style. I think others reworked it significantly later, > > > so if it's there at all it's worth looking at the earliest possible one. > > > > > > > The Pascal compiler rings a bell... It would be fun indeed to derived a > > version from it so that, finally, TeX and al. could be "natively" > > compiled instead of converting the (pseudo) Pascal to C (this is web2c > > purpose or, as I have named it, pp2rc---Pseudo Pascal to Raw C). > > There was an interpreter for P-code and (I think later) a compiler > for the Vax. You'd have to port it to current architectures, and > compiling TeX would probably make TeX run more slowly than the C version. > > The Berkeley Pascals were some of the compilers used for "Software Tools > in Pascal". The Pascal version would probably be a bit slower. And it would be more an alternative to verify the code than a primary way, since in fact D.E.K. has not written the program in some Pascal but in Algol, a high level abstract description, the wizardry being in the data structures. And, indeed, only the control flows are being translated from pseudo Pascal to C, the core---the data structures---being handled by ad-hoc code. And for the architectures, like other compilers, the aim would be to convert to some intermediate language (perhaps assembly) and to borrow the back-ends. But it is all vaporware: I have to implement supplementary primitives for the new version of LaTeX and I'm already late (not started yet)... -- Thierry Laronde http://www.kergis.com/ http://kertex.kergis.com/ http://www.sbfa.fr/ Key fingerprint = 0FF7 E906 FBAF FE95 FD89 250D 52B1 AE95 6006 F40C -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T476a1d7b83269775-Me5e205d80b9776ae87385457 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] APL
tlaro...@polynum.com wrote: > On Mon, Feb 22, 2021 at 09:44:54PM +, Charles Forsyth wrote: > > I'm fairly sure Thompson wrote it on sabbatical in Berkeley. I think he > > also wrote the first version of a Pascal compiler. > > Pascal isn't a difficult language but I remember that compiler having an > > unusual style. I think others reworked it significantly later, > > so if it's there at all it's worth looking at the earliest possible one. > > > > The Pascal compiler rings a bell... It would be fun indeed to derived a > version from it so that, finally, TeX and al. could be "natively" > compiled instead of converting the (pseudo) Pascal to C (this is web2c > purpose or, as I have named it, pp2rc---Pseudo Pascal to Raw C). There was an interpreter for P-code and (I think later) a compiler for the Vax. You'd have to port it to current architectures, and compiling TeX would probably make TeX run more slowly than the C version. The Berkeley Pascals were some of the compilers used for "Software Tools in Pascal". Arnold -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T476a1d7b83269775-M29f8ec6d6c8474c6d36e66e1 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
Re: [9fans] APL
On Mon, Feb 22, 2021 at 09:44:54PM +, Charles Forsyth wrote: > I'm fairly sure Thompson wrote it on sabbatical in Berkeley. I think he > also wrote the first version of a Pascal compiler. > Pascal isn't a difficult language but I remember that compiler having an > unusual style. I think others reworked it significantly later, > so if it's there at all it's worth looking at the earliest possible one. > The Pascal compiler rings a bell... It would be fun indeed to derived a version from it so that, finally, TeX and al. could be "natively" compiled instead of converting the (pseudo) Pascal to C (this is web2c purpose or, as I have named it, pp2rc---Pseudo Pascal to Raw C). -- Thierry Laronde http://www.kergis.com/ http://kertex.kergis.com/ http://www.sbfa.fr/ Key fingerprint = 0FF7 E906 FBAF FE95 FD89 250D 52B1 AE95 6006 F40C -- 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/T476a1d7b83269775-M3970db773e71081f039bfc81 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription