Introduce --lock resolves.
#1654
Conversation
For the local case of lock create and then later resolve, it will be good to retain the (primary) artifacts downloaded for the local interpreter when creating the lock used to lock in order to save downloading those artifacts again when later resolving from the lock. Work towards pex-tool#1583.
Resolving from a lockfile will need this too.
The Pex CLI can now resolve from a lock file created via `pex3 lock create` when creating PEX files. Closes pex-tool#1583
| @@ -77,6 +77,7 @@ def _add_resolve_options(cls, parser): | |||
| resolver_options.register( | |||
| cls._create_resolver_options_group(parser), | |||
| include_pex_repository=False, | |||
| include_lock=False, | |||
There was a problem hiding this comment.
@benjyw this is why it's an option to disable registration.
|
|
||
|
|
||
| class URLFetcherDownloadManager(DownloadManager): | ||
| _BUFFER_SIZE = 65536 |
There was a problem hiding this comment.
I hadn't seen this number used before but knew you chose it for a reason. Fine if you don't want to burn a tree, but otherwise this might help newbs like me
| _BUFFER_SIZE = 65536 | |
| _BUFFER_SIZE = 65536 # i.e. max number of 16 bits. |
Edit: or probably a better comment would be why this number.
There was a problem hiding this comment.
Yeah - not the reason. I can circle back, but, as always, buffer size selection is all about perf and nothing more. The number of bits may tangentially relate to the size of underlying os buffers, but its not the point, the point, hopefully, is you measured on some reasonably diverse set of inputs and found a best-fit. That's what this was.
|
Very exciting! Congratulations John. |
The Pex CLI can now resolve from a lock file created via
pex3 lock createwhen creating PEX files.Closes #1583