|The path to Go 1.1||Andrew Gerrand||3/5/13 3:50 PM|
Over the next month we should work toward preparing a release candidate, with the goal of releasing Go 1.1 RC1 during the week of the 1st of April.
At the beginning of February, Russ announced a design freeze: http://golang.org/s/go11freeze He said "new features or changes requested after today will be postponed to after Go 1.1."
I'd like to strengthen that message. From today, the only changes we - the Go contributors - should be writing and reviewing are those which address Go1.1-blocking issues. The list of blocking issues is here: http://swtch.com/~rsc/go11.html
This means that it's too late for optimization CLs, and way too late for feature proposals. Please shelve these until after the Go 1.1 release.
Thank you everyone for your tremendous efforts! The finish line is in sight.
|Re: The path to Go 1.1||lei zhao||3/5/13 7:03 PM|
在 2013年3月6日星期三UTC+8上午7时50分05秒，Andrew Gerrand写道：
|unk...@googlegroups.com||3/6/13 12:36 AM||<This message has been deleted.>|
|Re: The path to Go 1.1||supe...@gmail.com||3/6/13 9:33 PM|
|Re: The path to Go 1.1||jcna...@gmail.com||3/7/13 2:03 PM|
This is quite exciting, with all the improved stuff slated for 1.1! I checked out http://swtch.com/~rsc/go11.html# to see if I could lend a hand with anything. Looks like the easy stuff has been fixed already, but while looking, I did notice some issues
1). had a proposed fix but have fallen off the radar
2). are actually performance optimizations/new features that maybe shouldn't be labeled Go 1.1 at this point, in light of what Andrew Gerrand said below.
3). were considered too large to deal with shortly before Go 1.0.
4). are not clear (to me, at least) as to what the issue is.
I followed all the links and compiled a list below of issues that fall into one of these categories, with the hope that this will prevent confusion when others come here looking for things to work on. And if my efforts help core developers with bookkeeping, great -- if not, sorry for the noise...
L 3679 cmd/gc, time: make Format 5x faster
L 4438 cmd/gc: large unnamed struct can result in 'name too long' error at link time
-Doesn't say how large the struct was. Unlikely to occur in the field?
L 3506 cmd/go: (spurious?) rebuild due to compiler mtime
-With the workaround, is it still a release blocker?
L 3749 cmd/go: packages in different GOPATH are not rebuilt when modified
-This seems to have been downgraded to a doc fix. Is the 'L' designation intended for the doc fix or the larger issue?
M 2063 cmd/godoc: link identifiers to their declarations
M 3579 cmd/godoc: simplify documenting commands
-Should this be merged with 3199, as was suggested? That one is not a release blocker, and looks difficult.
M 4381 cmd/gofmt: incorrect comment placement for a comment following a function
-There doesn't seem to be agreement that this is a real bug, much less a release blocker.
M 3363 doc: ref/gdb should mention compiler flags / go install -gcflags
-Has this been fixed already?
L 2771 encoding/xml: MarshalXML interface is not good enough
-This seems to be a new feature/API, and no agreement on how to proceed.
L 3526 encoding/xml: name space ignored on attribute
-This seems to have stalled.
L 2362 image/jpeg: chroma downsampling ratios are restricted
-Long term issue, maybe not 1.1 blocker?
4475 misc/emacs: emacs gofmt error
-Is this still open?
L 4501 net: decide whether Zone makes the cut for Go 1.1
-Unlikely to make the cut at this point?
☞ M 3359 net/http: Documentation of implementation details for clients
-Was too late for Go 1 a year ago, too late again?
☞ M 4385 net/http: Redirect unresponding for URL containing Unicode
-There seems to be a way forward on this, but work has stalled.
M 2735 net/http/httputil: remove hop-by-hop headers in ReverseProxy?
-Is this still an issue?
☞ M 3594 net/http/httputil: add sort by quality
-New feature? Also, last update was a month ago.
4687 net/mail: Subject header is not RFC2047-decoded
-The discussion seems to indicate an API change might be needed for this. Is that OK at this point?
4700 net/url: Parse("#foo") should not result in error
-There seems to be a CL, but no discussion.
4706 net/url: better path resolving
-Seems to have stalled.
☞ L 3358 os: some functions fail if full paths are long (more than about 250 chars) on windows
-Was too late for Go 1 last year, too late again?
☞ M 4290 os/exec: Wait should not close client side of pipe [started by rog...@gmail.com]
-This seems to have been downgraded to a doc fix, which seems to have stalled.
4068 os/fsnotify: add new package
M 2320 reflect: Type.Field allocates [started by buil...@golang.org]
-Unsure if this is intended to be a performance optimization, but in any case was punted last year, punt again?
M 4174 runtime: Func.FileLine doesn't handle multiple methods on a single line correctly [started by ]
-Seems to have stalled.
L 3751 strings: Index(s, "x") not as optimized as bytes.IndexByte
L 1435 syscall: Setuid/Setgid doesn't apply to all threads on Linux
-This is a long-term issue that seems to require an API change, and perhaps invasive changes in the runtime.
☞ L 2981 testing: add -json flag for json results
4553 time: update packaged time zone information for Go 1.1
-Is it close enough to worry about, yet?
4918 website: associate specific color with examples and runnable code
-Not sure why this is Go 1.1.
|Re: [golang-dev] Re: The path to Go 1.1||Dan Kortschak||3/7/13 2:51 PM|
This is one that I am interested in, and have looked for discussion
which might suggest a direction, but this seems lacking. In its absence
I have put together a CL that (although not 100% passing) could be a
start to the discussion. Unfortunately it has receiver very little