441500+ entries in 0.283s
ascii_field: but i like
the idea of owning a lifetime's supply of non-nato cpu.
ascii_field: trinque: what is interesting
to one is just old junk
to another.
trinque: ascii_field: might be cool
to get a rough list of interesting historic artifacts from you sometime
ascii_field: not 5 years from now when
the fakes are stamped out by
the millions.
ascii_field: other
thing is, items like z80 or rom of ~known vintage~ can be considered 'verifiable' in a weaker sense of
the word
punkman: ascii_field: solder, sure. what's
the most complicated component we can verify in jungle? (is
there previous
thread?)
nubbins`: turns out it's an issue with
the cartridge in
the CIS kit
ascii_field idly wonders how big z80 would be if
turned into discrete
ttl
ascii_field: in
that sense jurov is right, all microprocessors ever made are 'closed source'
assbot: Logged on 12-07-2015 22:31:52; jurov: but anyway, what i'm interested in,
that maybe really 10nm fab is not needed and lesser process would suffice
ascii_field: for all we know, usg just finished installing 900Mhz intercept network and now is banking on 'streisand effect'
to get us
to use it.
ascii_field: punkman: not like usg considers
the question 'is
this
trivial
to make' before
they gag
punkman: "Designed
to augment existing privacy
tools, ProxyHam is a Raspberry Pi computer with Wi-Fi enabled." << gag order for
this??
ascii_field: perhaps
the ultimate joke. yet another schmuck helping
to redefine what 'open' means,
to
the peanut gallery
jurov: z80 is closed source
turd,
too, no?
assbot: Logged on 12-07-2015 22:24:39; *: jurov expects asciilifeform
to shoot holes in it ^
ascii_field: 'it appears
that his hand was forced – legally – complete with gags and destruction orders.'
ascii_field: 'Effective immediately, we are halting further dev on #proxyham and will not be releasing any further details or source for
the device... Existing #proxyham units will be disposed of and no longer be made available at @_defcon_'
ascii_field: ^ seems
to be a modernization of
the ancient 'gold box'
trinque: yet
they called it a "secure web interface"
ascii_field: 'Trend Micro
Threat Intelligence Manager installs a secure web interface (
httpd.exe, tcp port 443/
https) which listens for incoming requests. Several vulnerabilities have been found in
the product
that would allow a remote attacker
to cause
the product
to execute arbitrary code.'
phf: ascii_field: i don't
think
that's
the case. i changed block processing
to return false when
the block is bastard, and eatblock started returning false for
those.
ascii_field: prolly
the innumerable 'critical section' nonsense
ascii_field: or say i wanna pipe my node's log
to a
thermal printer.
ascii_field: would be neat if i could learn
this from looking at
timestamps
ascii_field: ben_vulpes: say, figuring out how long it
takes
to verify block N
ben_vulpes: punkman: i'm not advocating insane logs, i'm just saying pretty it up and make it consistent but perhaps we don't waste
too much
time arguing about syslog and logrotation at precisely
this moment?
trinque: having watched many movies on
the subject
trinque: "But if so, he'd be
the least well hidden operative ever." << lol
punkman: ben_vulpes: I disagree, how
the hell are we supposed
to debug with insane logs?
ben_vulpes: also logging is a pretty low priority
ticket. if it looks better, has more information, is less braindamaged, is more consistent, great. does not deserve major surgery atm imho.
trinque: you could have
the best of both by having it just fart messages out stdout and stderr,
then attaching it
to something like svlogd if/when you want
that in a file
ben_vulpes: can we get
the best of both worlds? eg log
to standard debug.log if unconfigured, syslog if configured?
ascii_field: i.e. one more
thing i'd have
to reimplement for msdos or bare-metal port.
ascii_field: again
this is entirely logical except
that it creates a new dependency
phf: not running any syslog daemon
turns it into fire and forget, so you won't get any logs unless you are physically connected
to console
phf: ascii_field: autorotators exist because people don't know how
to syslog mostly. i.e. you build a syslog demon based on
the constraints of your system, i.e. on pogo might just be busybox's syslog,
that does rudimentary
truncation. elsewhere syslog writes
to postgresql or whatever
trinque: (I'm not aware of whether it already has
this)
ben_vulpes: is disabling logs on
the pogo not an adequate solution for you?
trinque: what about a signal for it
to close and reopen
the log, and
then people can do whatever
they please outside it?
ascii_field: i also dislike increasing
the subset of libc we use
ascii_field: ben_vulpes:
thing is, i like cutting code. but i dislike inserting dependencies on complicated unixisms like autorotators.
ben_vulpes: ascii_field:
turn it off on
the pogo.
they're fire and forget, are
they not?
trinque: I'm sure my cats would have
things
to say about fair if I stopped feeding
them.
ben_vulpes: great big screaming steamer of "i cannot reason about
this system"
nubbins`: jesus
that reddit diatribe was rough
ascii_field: ben_vulpes: i'm all ears re: an alternative
that doesn't
turn my 5MB pogo rom into 10
punkman: ben_vulpes: agreed, I'd like
to make
the log more parsable
ben_vulpes: ascii_field:
the debug.log
thing is horrendous imho
nubbins`: fair generally means "people have vastly different quantity and quality of
things"
nubbins`: but
that's rarely, if ever, fair
nubbins`: people seem
to
think fair means "everyone has
the same
things"
phf: but i
think
the easiest option is
to adopt syslog interface (since
they already have a reasonable list of priority
tags), make it go
to debug.log by default, and
then interested parties can just do #define log syslog
phf: i
think libc's syslog facility
tries
to pass messages
to syslog and if
the daemon is not running,
they go into /dev/console instead
ascii_field: but keep in mind
that
the
thing needs
to work on machines with no perl, no bash, etc
ascii_field: and log rotation, etc. ought
to happen in separate proggies
ascii_field: i'll let others comment. maybe i'm wrong about
this one
ascii_field: phf: i'm biased in
the direction of rolling functionality ~into~ bitcoind, rather
than ~out~
to unixland
phf: ascii_field: we seem
to agree on loglevels, but what's your
take on feeding logs into syslog?
assbot: Logged on 13-07-2015 14:42:55; punkman: if you wanna logrotate,
there's logrotate?
mod6: ben_vulpes, ascii_field, et. al. anyone have any other
thoughts on
this ^^
mod6: i see
that now.
that's fine
then.
phf: it might be worthwhile
to replace all
those printfs with syslog(3) calls, specifically introducing
the priority as
the first argument, i.e. LOG_ALERT, LOG_WARNING, LOG_DEBUG. maybe have it conditional compilation, so
that real bitcoin instance will log
to syslog, where's debug instance will still write
to file
mod6: phf: yah was just about
to say, maybe wait until he submits his,
then you can apply yours on
top.
punkman: punkman:
the change from calling snprintf
to calling my_snprintf <<
there was a "#define snprintf my_snprintf" before, which I removed
phf: mod6: ah good
to know. i'll resubmit, not sure how my patch interacts with recent changes from punkman yet
punkman: yeah forgot
to remove
that, will do
mod6: but if we /always/ want
them in
there, maybe we should remove
the commandline flag.
mod6: punkman: my
thoughts
too.
ascii_field: thestringpuller:
the obsession with 'fair' is interesting - psychiatrically
punkman: mod6: what kind of insane person doesn't
timestamp logs?
assbot: Logged on 13-07-2015 14:42:16; asciilifeform: and what's
the deal with removing ShrinkDebugFile ?
mod6: thanks for putting effort into cleaning
this stuff up.
mod6: so other
than what I said above about keeping FormatException
the same, and leaving fLogTimestamps = false, and asciilifeforms' notes above, looks pretty good :]
mod6: why are boost/function.hpp and wallet.h added
to headers.h ?
mod6: punkman: also,
there is a commandline option for -logtimestamps so maybe just leave fLogTimestamps = false;
mod6: punkman:
the change from calling snprintf
to calling my_snprintf (which inturn calls vsnprintf) might be a good change. but unless we know why we're doing it, maybe we should just leave
that alone.