npub1dj7q4kzehyveqtxhpmrgjpclxw0sfuuy9rqazutl4pfch3vmuurq6txuj8 (npub1dj7…xuj8) that's why my proposal is along the line of overriding thread-unsafe functions with `*_safe` twins - it wouldn't break back-compatibility, and it'll introduce changes (veeery) gradually into codebases as the new standard is adopted. Implementing a reentrant multilock mechanism that simply uses memory locations as keys wouldn't even be technically difficult to implement.
But I've got the impression that POSIX will simply not easily accept such a big change, no matter how well-intentioned or well-designed. POSIX basically hasn't changed at all since 1997 - the three reviews that occurred since then, in 2001, 2008 and 2017, have all been very minor and usually amounted to errata or minor fixes. Things can be designed to be back-compatible, but old standards that are so widely used are just insanely resistant to change...
Fabio Manganiello
npub1v7…ukv0u
2023-11-20 11:37:56 UTC
in reply to nevent1q…6pag
Author Public Key
npub1v78mmuz20p6qd6nve30axhqu74avwn4f6z4grhug7755rat7wh3syukv0uPublished at
2023-11-20 11:37:56 UTCEvent JSON
{
"id": "ce2b96be5837535b5f4a421339e4f43aae55f3938053a4cdcd3dd6ec6b78a722",
"pubkey": "678fbdf04a787406ea6ccc5fd35c1cf57ac74ea9d0aa81df88f7a941f57e75e3",
"created_at": 1700480276,
"kind": 1,
"tags": [
[
"p",
"6cbc0ad859b919902cd70ec689071f339f04f38428c1d1717fa8538bc59be706",
"wss://relay.mostr.pub"
],
[
"p",
"1b00bfa5e877706dedbd14c55acf7eeb5de3a768137b8c557215dd91f3335ebf",
"wss://relay.mostr.pub"
],
[
"e",
"457fb1db06ca4e76918b1020e04a63b5aec9fecbf09b22269b74fa8dcc5c2de9",
"wss://relay.mostr.pub",
"reply"
],
[
"proxy",
"https://social.platypush.tech/users/blacklight/statuses/111442675377428797",
"activitypub"
]
],
"content": "nostr:npub1dj7q4kzehyveqtxhpmrgjpclxw0sfuuy9rqazutl4pfch3vmuurq6txuj8 that's why my proposal is along the line of overriding thread-unsafe functions with `*_safe` twins - it wouldn't break back-compatibility, and it'll introduce changes (veeery) gradually into codebases as the new standard is adopted. Implementing a reentrant multilock mechanism that simply uses memory locations as keys wouldn't even be technically difficult to implement.\n\nBut I've got the impression that POSIX will simply not easily accept such a big change, no matter how well-intentioned or well-designed. POSIX basically hasn't changed at all since 1997 - the three reviews that occurred since then, in 2001, 2008 and 2017, have all been very minor and usually amounted to errata or minor fixes. Things can be designed to be back-compatible, but old standards that are so widely used are just insanely resistant to change...",
"sig": "87e24079bb0ea90ec5d8a41f9ab49ec5f4b2193a2c1d5617b82d3c985ce0c92dbe1ba48bf1a0270a9a7a095f950e2d46fd9711a8e9a86b8dda5a3a9649af9e93"
}