Hide Idle (>14 d.) Chans
← 2023-07-05 | 2023-07-07 → ↓
signpost[asciilifeform]: yup, one concern was just being able to cheaply ignore.
awt[asciilifeform]: !!ticker btc usd
awt_akris[asciilifeform]: Yes an option to keep or discard a binary would be good. Currently can't see how to do that without having gaps in the chain.
awt_akris[asciilifeform]: Which brings up the question, should operators be able to delete *any* packet they do not wish to keep or serve?
awt_akris[asciilifeform]: !!ticker btc usd
asciilifeform: awt_akris: broken chain will rather obv prevent a peer from getdataing what came prior to the break
asciilifeform: hence the thrds re 'do you really want multi-MB lolcats in the chain'.
awt_akris[asciilifeform]: yes, asciilifeform, that is obvious
asciilifeform proposed an 'inv' to drywall over this problem but imho it is a questionable pill
asciilifeform: incidentally, even without 'i want to delete $msg', you still gotta have some means for storing hashes of msgs you don't want the station to eternally try getdataing, because by all indications they simply don't exist (i.e. chain hashes produced by buggy clients)
asciilifeform suspects that there's ultimately no way to avoid introducing some alt-mechanism for fetching old msgs that doesn't rely on an intact chain
awt_akris[asciilifeform]: Isn't this just select * from asked_for where attempts > n?
asciilifeform: awt_akris: aha
asciilifeform: 'GetDataTries' in curr. draft spec
awt_akris[asciilifeform]: What about a "placeholder" packet that says "I don't have this *anymore*, but maybe I have this packet that came immediately before, here's the hash"
asciilifeform: awt_akris: this was asciilifeform's notion behind 'inv'
asciilifeform: sect 5.4.9 in draft
awt_akris[asciilifeform]: Yeah I see it. How would an inv be populated? Randomly?
asciilifeform: awt_akris: asciilifeform did not specify in draft. 1 possible way is what awt_akris described above -- bridging a broken chain
asciilifeform: 'i haven't got $msg but i do have these-here with earlier timestamps than anyffin you've asked for'
asciilifeform: the other notion behind 'inv' was the fact (elaborated at length by phf) that if netchain ends up used for 'threading', structure of the msgs will be a tree rather than linear chain, and there oughta be a way for getdataing station to retrieve the subtrees
awt_akris[asciilifeform]: Nice. Got a binary pacakage working on linux.
awt_akris[asciilifeform]: via pyinstaller.
unpx[asciilifeform]: http://logs.nosuchlabs.com/log/pest/2023-07-02 <<< I'm still catching up. Re vpatches, and sharing tools: I understand phf's point on ``vpatches' idea'', but it is also clear that signpost is trying to remove sadism from a good thing. Exactly like I can distribute a tarball for personal purposes, anyone can just trust my s
unpx[asciilifeform]: ignature and install. Indeed, these are either robots or monkeys.
signpost[asciilifeform]: there was this lingering conceit that one'd only express fits-in-head items in vpatches. this resulted in carrying along wads of e.g. ave1's scraped together version of gregor-somebody's gcc+musl bootstrap.
signpost[asciilifeform] fully supports patch reading, just doesn't think pretending these items are outside the system helps.
signpost[asciilifeform]: and I don't mean anyone in particular when I say this, but an mp character screeching at untermenschen chills the air to the degree that all behavior nervously avoids the rebuke.
signpost[asciilifeform]: at any rate, if one wants to read some patches, the patches that actually comprise an e.g. built trb are there, in pentacle's /src/ dir.
signpost[asciilifeform]: vtools also. it takes a gnat. this is not avoidable at the present time. that doesn't mean I like it.
unpx[asciilifeform]: I have two eyes but I can read one line at a time for now
unpx[asciilifeform]: I think someone need to think why is using vtools, sane editors, readable code and signed patches necessary. My knowledge on ADA or CL is useless at the moment and won't be able to fill this hole now. I'm thinking about what I need, and since I can't build everything on my own, even if I had the time, I think I should focu
unpx[asciilifeform]: s on tools that actually matter to me (e.g. not rewriting a terminal emulator)
unpx[asciilifeform]: One may even write a tool that tries to apply every patch by bruteforce
unpx[asciilifeform]: One one hand one may want to write her own birdfeeder, on the other hand one shouldn't reinvent every wheel. I don't see why non-portable code is written, like everything will run on Gentoo forever. To me it would be better to have blueprints so I can write a working piece in a weekend
signpost[asciilifeform]: is this chatgpt?
signpost[asciilifeform] scratches head, lol
unpx[asciilifeform]: SSH is slow now, sigh
signpost[asciilifeform] doesn't follow is all.
unpx[asciilifeform]: signpost: that thread, and others, are not just about vpatches tbh
unpx[asciilifeform]: they are about the ``religion'' you are asking to review, the one where each one is able to work everything on their computers and the only thing that matter is trb's patches
signpost[asciilifeform]: sure, 100% the case.
unpx[asciilifeform]: Please remove sarcasm
signpost[asciilifeform]: there wasn't any.
signpost[asciilifeform]: it's 100% the case.
signpost[asciilifeform] has routinely found that what reads to him like flat text is read by others as screaming.
signpost[asciilifeform]: or sneering, what have you.
signpost[asciilifeform]: at any rate, yep, a (mostly simulated) physical ritual of downloading a patch is not what makes one the kind of person that reads the patch.
signpost[asciilifeform]: it's that thing I was trying to do - and apparently failed to achieve - which is debate the method of uncovering truth in the forum.
unpx[asciilifeform]: To me is the fact that the whole process is not clear, indeed why have a vtool when I can review and apply manually patches?
shinohai[asciilifeform]: I should write a bash script that downloads patches and forces you to read them piped through `less` for added masochism.
unpx[asciilifeform]: shinohai, if you really want to, you can, the point is you are not explaining me a lot of things I may need to be able to review it
unpx[asciilifeform]: Like I know some languages (e.g. C, Scheme, Python) and understand a bit CL, but I would first need to read a reference manual on ADA before understaing even the gnat build configuration
unpx[asciilifeform]: So what is best? Giving me the vtools in a tarball, giving me the vpatches and letting me suffer the unknown or giving me the blueprints and the vpatches so I may choose to write my own?
signpost[asciilifeform]: the upstream thing is what's wanted, and who says.
unpx[asciilifeform]: I think there should be a way to manage documentation, not just a link to an archived site. That's why I was thinking how to manage everything in a more sane way: docs, code, vpatches, viewer, editor and environment
unpx[asciilifeform]: Or at least guide people so they can write they own tools to do all of that the way they want
signpost[asciilifeform]: these tools were originally conceived in the context of the tmsr concept of sovereignty, which to quote the declaration held "nothing above, and the whole world below"
signpost[asciilifeform]: out of that individual sovereignty emerges as a necessary consequence self-reliance in thought, in business, in being-in-the-world.
signpost[asciilifeform]: this is all beautiful, and the practical reality is we're years away from the kind of technogenic hell the chinese live within.
signpost[asciilifeform]: *only years away
signpost[asciilifeform]: we being whatever the pitiful "west" is now.
dulapbot: (trilema) 2015-08-05 trinque: I had a hell of a time figuring out where I should be re: patches, at least before giving up and using rotor instead.
dulapbot: (trilema) 2015-08-05 jurov: but srs, i have asked for patches context, was sent to go scour the log for it
dulapbot: (trilema) 2015-08-05 asciilifeform: as i understand, any mechanized thing really depends on patches stating their antecedents
asciilifeform: /log/trilema/2015-08-06#1228194][and]
bitbot[asciilifeform]: Logged on 2023-07-06 18:52:42 unpx[jonsykkel|deedbot|signpost]: To me is the fact that the whole process is not clear, indeed why have a vtool when I can review and apply manually patches?
bitbot[asciilifeform]: Logged on 2023-07-02 21:38:41 asciilifeform[4]: recall that asciilifeform originally advocated hand-sewn equivalent of this process before wrote vtron
dulapbot: (trilema) 2015-08-06 ascii_field: ~verifying~ left - for now - to the alert reader.
asciilifeform: http://logs.bitdash.io/pest/2023-07-06#1027951 << ffa series ch.1 has 100% of errything one'd want to know re 'how do i bake builds on gnat'
bitbot[asciilifeform]: Logged on 2023-07-06 18:56:05 unpx[jonsykkel]: Like I know some languages (e.g. C, Scheme, Python) and understand a bit CL, but I would first need to read a reference manual on ADA before understaing even the gnat build configuration
asciilifeform: worked example.
signpost[asciilifeform]: https://www.adacore.com/download/more << adacore does not offer downloads earlier than 2018 anymore
signpost[asciilifeform]: waitaminute, I suck cocks
signpost[asciilifeform]: had the wrong arch selected
dulapbot: Logged on 2022-11-06 23:08:05 asciilifeform: asciilifeform's mirror of adacorpse's gnat fwiw
signpost[asciilifeform]: if y'all don't have any use for my item, that's alright by me. it really doesn't affect me.
signpost[asciilifeform]: I was just indicating I'm aware of the philosophical roots of the item under discussion.
asciilifeform not tested signpost's latest just yet, but intends to
asciilifeform: review of roots was for unpx
asciilifeform: ( signpost was obv there & saw how & why we have vtron )
signpost[asciilifeform]: one place I would like to improve the captured gcc specimen is 100% reproducible builds. cranked it pretty close, but not all the way yet.
signpost[asciilifeform]: that was one of my primary motivations for capturing.
asciilifeform: signpost: what was the remaining source of irreproducibility ? ( was it the timestamps-in-the-binary idjicy ? )
signpost[asciilifeform]: squashed that one with __DATE__ and __TIME__ and making sources all have the same file attr timestamps.
asciilifeform: a hm, what remained then ?
signpost[asciilifeform]: there were a few other sources of churn I squashed, but didn't discover all of them yet.
awt_mbpro[asciilifeform]: On osx the binary build looks like this: http://paste.deedbot.org/?id=_8uG. Using it now.
← 2023-07-05 | 2023-07-07 → ↑
Random(pest) | Download hourly DB snapshot | Get Source Code