Newsletters: add #9 (2018-08-21) #58
Conversation
|
There was an error in last week's newsletter:
bitcoin/bitcoin#13925 was a merge of bitcoin-core/leveldb-subtree#19, which increased the max number of mmaps, not the max number of file descriptors. How do people feel about an errata section? (thanks to @sdaftuar for pointing this out) |
|
Both mmap(2) and select(2) make file descriptors available per the docs and (IIUC) ldb was falling back on select when it hit its internal maximum limit on mmap. By increasing ldb's internal mmap max, it can now open more file descriptors than before (still falling back on select, although that should be unnecessary now for a bunch more years hopefully). So I think it's correct to say, "increases the maximum number of file descriptors Bitcoin Core's internal database can use". The consequence of this change is that ldb shouldn't ever call select during normal operation and won't use any part of select's glibc-imposed per-process maximum fd count, allowing other parts of bitcoind that are designed around select to make use of that capacity. So I also think it's correct to say, "which can allow more file descriptors to be used for network connections." Even if the correct way to phrase that was to say "more mmaps" or something, I'd suggest just editing the archive copy of the newsletter rather than printing a correction as long as the conclusion is correct. I doubt anyone besides glibc cares how we open files. :-) (But if the conclusion is in error, I'm more than willing to write the Corrections section myself.) |
|
Thanks for the excellent explanation David! I don't think there's any need to update newsletter 8. |
|
I'm on vacation this week + don't have anything at the ready for dashboard section. I suggest we a) omit or b) add standard "low fees" bullet or c) if someone wants to go find an actionable nugget, go for it. |
|
ACK |
| major release such as this, but each RC can theoretically be the last | ||
| RC, so we encourage you to test as early as possible. | ||
|
|
||
| ## Dashboard items |
There was a problem hiding this comment.
No dashboard this week. Remove!
| layout: newsletter | ||
| lang: en | ||
| --- | ||
| This week's newsletter includes the usual dashboard and action items, |
There was a problem hiding this comment.
No dashboard this week. Remove "dashboard and ".
| - **Output script descriptors enhancements** being worked on by | ||
| Pieter Wuille. The basic idea for this was described in | ||
| [Newsletter #5][news5 news] but Wuille is investigating adding | ||
| support for the ability to "import `and(xpub/...,or(xpub/...,xpub/...))` |
There was a problem hiding this comment.
I think using the words "nested and/or/threshold constructions" as Pieter did in IRC is a clearer way to introduce this (but maybe that's just a matter of personal taste)
| measurements by Naumenko). Maxwell is also working on making it | ||
| possible for a newly-started (or long-disconnected) node to | ||
| efficiently sync its mempool with its peers using this same basic | ||
| mechanism, reducing the requirement for small independent miners to |
There was a problem hiding this comment.
I'd lean towards removing everything after "mechanism"
| percentile feerates for a historic block with the `getblockstats` RPC | ||
| introduced to the master development branch a couple months ago. | ||
|
|
||
| - [LND #1693][] allows LND's autopilot funding mechanism to optionally |
There was a problem hiding this comment.
As discussed, the primary motivation for this PR is to allow autopilot to open many channels in parallel, since it doesn't need to wait for one channel to be confirmed before using the change to open the next channel.
harding
force-pushed
the
2018-08-21-newsletter
branch
from
5da9cb3 to
321a501
Compare
|
ACK 321a501 |
jnewbery
force-pushed
the
2018-08-21-newsletter
branch
from
321a501 to
1abcbfe
Compare
3 participants
Note: I'll be AFK most of the day tomorrow (Monday), so please push changes without waiting to hear back from me. I'll review later in the evening (EDT) when I get home.