| 00:45 |
awt |
inheritance in python seems to be a total hack |
| |
~ 19 minutes ~ |
| 01:05 |
awt |
on the plus side pycharm probably cuts refactoring time in half by catching many of my syntax and other errors while I type |
| 01:15 |
awt |
I probably missed this in the spec - but say your station goes offline, you miss some messages, then comes back online and the speaker of the missed messages never again speaks. How do you get those missed messages? |
| |
~ 22 minutes ~ |
| 01:37 |
asciilifeform |
awt: broadcast msg? |
| 01:37 |
asciilifeform |
... then you getdata it from peers. |
| 01:38 |
asciilifeform |
( if was direct msg, it's lost 4evah, you won't even find out that it was sent ) |
| 01:45 |
awt |
asciilifeform: when would your peers send you the messing messages? |
| 01:45 |
asciilifeform |
awt: when you know it existed (have a chain that points to it, either from a successfully received broadcast msg, or from a prod) and getdata for it |
| 01:46 |
awt |
ahah - prod. ok |
| 01:46 |
asciilifeform |
*subsequent broadcast msg |
| 01:46 |
asciilifeform |
i.e. when you learn you have a chain gap (exactly 2 ways to learn this) |
| |
~ 13 hours 3 minutes ~ |
| 14:50 |
awt |
asciilifeform: broken link in the draft: http://www.loper-os.org/pub/pest/33161-broadcast-text |
| 14:51 |
asciilifeform |
and oh hm billymg's logger fell down ? |
| |
↖ |
| 14:51 |
asciilifeform |
awt: indeed broken. (would be a++ if had a mechanical detector for these...) |
| |
↖ |
| |
~ 1 hours 13 minutes ~ |
| 16:05 |
billymg |
http://logs.bitdash.io/pest/2022-02-02#1003477 << looks up still |
| 16:05 |
bitbot |
Logged on 2022-02-02 14:51:02 asciilifeform: and oh hm billymg's logger fell down ? |
| |
~ 18 minutes ~ |
| 16:23 |
asciilifeform |
billymg: hm interesting, must've been stalled somehow earlier |
| 16:24 |
billymg |
yeah, it's laggy sometimes |
| |
~ 1 hours 1 minutes ~ |
| 17:26 |
awt |
http://logs.bitdash.io/pest/2022-02-02#1003478 << I used to use this thingy on my mb: https://peacockmedia.software/mac/integrity/free.html (or something like it) |
| 17:26 |
bitbot |
Logged on 2022-02-02 14:51:20 asciilifeform: awt: indeed broken. (would be a++ if had a mechanical detector for these...) |
| 17:32 |
awt |
asciilifeform: http://www.loper-os.org/pub/pest/pest_draft.html#322-selfchain - regarding "skips" here, what does "chronologically" have to do with it? Isn't any chain forked with two messages with the same handle and selfchain regardless of if their timestamps are identical or not? |
| 17:38 |
asciilifeform |
awt: indeed oughta be chain-preceding, not chronological |
| 17:39 |
* |
asciilifeform plans to rewrite all of the pertinent sections in light of new algo |
| 17:39 |
* |
asciilifeform sadly hands 100% full atm |
| 17:43 |
awt |
asciilifeform: please reply at your leisure. Just working through this with pen and epaper. |
| 17:44 |
asciilifeform |
awt: simply must note, the draft spec currently not includes any of the items in linked thrd |
| 17:47 |
awt |
asciilifeform: noted |
| 17:52 |
awt |
asciilifeform: When a n00b joins a net, will all messages from the genesis message be dumped in his console? Or perhaps only the the last hour's worth, the rest going directly to cold storage? |
| |
~ 18 minutes ~ |
| 18:11 |
asciilifeform |
awt: good q. did not specify this |
| 18:11 |
asciilifeform |
(given as it aint pertinent to protocol function; tho similar argument can be made for several mechanisms which did specify) |
| 18:12 |
asciilifeform |
in a hypothetical 100% ab initio (i.e. non-irctronic) pestron, would make sense to 'scrollback' for old msgs as in certain heathen chats |
| 18:12 |
asciilifeform |
for irc, prolly does not make sense to dump potentially GB+ of text into a console, no |
| 18:12 |
asciilifeform |
rly oughta be a knob |
| |
~ 21 minutes ~ |
| 18:34 |
awt |
http://logs.nosuchlabs.com/log/asciilifeform/2022-01-26#1076762 << here the chain number indicates the order by timestamp of the message subsequent to the last message of the unforked chain (seems like there must be better terminology for this)? |
| |
~ 2 hours 2 minutes ~ |
| 20:36 |
asciilifeform |
awt: oughta snip all 'order by timestamp', yes |
| 20:36 |
asciilifeform |
there's no place where any such thing needed, when you've the chains. |
| 20:36 |
asciilifeform |
the only thing ordered by time (of receipt, that is, rather than timestamp) is the holding buffers |
| 20:37 |
awt |
asciilifeform: in that case what determines the order of chain numbers? |
| 20:38 |
asciilifeform |
hm good q |
| 20:38 |
asciilifeform |
rly oughta be position on chain. |
| 20:39 |
asciilifeform |
if we prohibit output to console of broken chains , can be consistent/immutable. |
| 20:41 |
awt |
asciilifeform: if two nodes on the chain have the same parent, aren't the positions the same? |
| 20:43 |
PeterL |
was it going to output to console after some time if the chain was never patched? |
| 20:44 |
PeterL |
or just drop broken chains after some interval? |
| |
~ 28 minutes ~ |
| 21:12 |
asciilifeform |
orig. the former. imho the latter is The Right Thing |
| 21:13 |
asciilifeform |
awt: thinking about it, yes. needs some canonical way to enumerate. lemme know if you think of one. |