<<  >> (p.1482)
    Author Topic: Gold collapsing. Bitcoin UP.  (Read 2032327 times)
    solex
    Legendary
    *
    Offline Offline

    Activity: 1078
    Merit: 1007


    100 satoshis -> ISO code


    View Profile
    August 01, 2015, 11:17:07 PM
     #29621

    Unfortunately, eliminating the limit would open up the existing code bases to several of DOS attacks. If we really want to go that way, we need to make sure that all the software can handle blocks that won't fit in memory + swap.  There are also likely other technical issues that would need to be addressed, and the only way to tease them all out is a ton of code reading and testing.  When software is written, constants are treated as assumptions, and when that assumption changes there is a risk of the previous logic breaking down.

    The tcp buffer limit is 33554432 bytes (33.5MB) which means that Bitcoin messages cannot exceed this or p2p transmission breaks at the receiver with a read error.
    If the 1MB limit did not exist then this is the maximum block size that can be sent, and it is no good mining a block too large to send anywhere. Blocks not fitting in memory or swap is not the immediate restriction.

    Jeff Garzik commented that increasing this limit means a major reworking of Bitcoin's p2p protocol. If anyone thinks that the 1MB hardfork is scary (which is a tiny software change) resolving the 33.5MB is a much bigger issue. It may never be done, because one or more of alternative options:

    LN having enough block space to allow a *lot* of payment channels to function and the VISA-scale volumes do-able long-term.
    Tier Nolan Adam Back extension blocks are one solution when the 33.5MB limit is being approached,
    A combination of IBLT (if it is efficient at this level) and block segmentation logic could allow disk blocks > 33.5MB

    In the meantime (and I'm guilty in the past too) bitcointalk is full of comments about 100MB, ZOMG 1GB blocks, etc etc and "dangerous" unlimited block sizes, when the reality is that no block can exist > 33.5 MB even if the 1MB limit disappeared.

    When Satoshi put in the 1MB it was 1000x larger than the average block size. The 33.5 is about 80x the average block size seen in 2015. This is large, but not stupidly large, and continuing improvements in computing tech and bandwidth should make this size manageable by the mid-2020s because 5MB blocks should be manageable today.

    A road-map which permits 33.5MB blocks by about 2025 still allows Bitcoin to be a major global currency. BIPs 100 and 101 at least leave the opportunity open to make this possible.

Page 1481
Viewing Page: 1482