You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Fix main to not take args, just have a sys::get_argv or something
Fix the Eq and Cmp traits
Add a -Z flag for compiling with the old mode logic; turn old mode logic off by default, only on for rustc (and/or old libraries if they don't get done in time)
I prefer to keep the main argv signature, mostly because sometimes it's useful to call main manually, and not having an argv in the signature means involving global runtime state. For example, the jit is currently tested by building crates and calling main.
I agree with @brson. We were discussing this the other day and I remembered how much I dislike how Python does it. I always end up making a main(args) fn anyhow. Basically it's bad for all the normal reasons global state is bad—it's unclear what the inputs to the fn are, and it gives the caller no control over them.
Current status: serialization and serialization2 cause several mode warnings, and net_tcp causes one warning. Two others (in ptr and extfmt) will go away once I make a new snapshot. This is otherwise done.
Still warnings from (in core) stackwalk::frame_address, and (in std) net_tcp and serialization. Otherwise done. Deprecated modes forbidden in all modules except these, AFAIK.
Activity
graydon commentedon Aug 31, 2012
Some other components of this bug:
main
to not take args, just have asys::get_argv
or somethingEq
andCmp
traitsbblum commentedon Sep 4, 2012
why not have main's optional arguments be
&[~str]
instead of&&~[~str]
?jdm commentedon Sep 5, 2012
One benefit of a sys::get_argv is that it reduces the "public static void main(String[] args)" factor.
brson commentedon Sep 29, 2012
I prefer to keep the main argv signature, mostly because sometimes it's useful to call main manually, and not having an argv in the signature means involving global runtime state. For example, the jit is currently tested by building crates and calling main.
nikomatsakis commentedon Sep 29, 2012
I agree with @brson. We were discussing this the other day and I remembered how much I dislike how Python does it. I always end up making a
main(args)
fn anyhow. Basically it's bad for all the normal reasons global state is bad—it's unclear what the inputs to the fn are, and it gives the caller no control over them.catamorphism commentedon Oct 1, 2012
Current status:
serialization
andserialization2
cause several mode warnings, andnet_tcp
causes one warning. Two others (inptr
andextfmt
) will go away once I make a new snapshot. This is otherwise done.catamorphism commentedon Oct 5, 2012
Still warnings from (in
core
)stackwalk::frame_address
, and (instd
)net_tcp
andserialization
. Otherwise done. Deprecated modes forbidden in all modules except these, AFAIK.catamorphism commentedon Oct 6, 2012
I'm closing this because we've done all we're going to do for 0.4; the ones I mentioned in my previous comment have been muted.
Auto merge of rust-lang#5536 - rail-rain:fix_manual_memcpy, r=phansch
Merge pull request rust-lang#3323 from scampi/failure-repo
Fix visibility of some Kani intrinsics (rust-lang#3323)