Rob Pike [Sun, 19 Sep 2010 21:37:06 +0000 (07:37 +1000)]
gob: make robust when decoding a struct with non-struct data.
The decoder was crashing when handling an rpc that expected
a struct but was delivered something else. This diagnoses the
problem. The other direction (expecting non-struct but getting
one) was already handled.
Robert Griesemer [Thu, 16 Sep 2010 20:45:40 +0000 (13:45 -0700)]
godoc: show "Last update" info for directory listings.
Use -timestamps=false flag to disable.
(This used to be shown on the front-page below the
build information with the old godoc. However, the
time stamps are directory-specific and should be
shown with the directory.)
* SHOW DB responds with a
"110 n databases present - text follows" -- parse it.
* Doing a Define() on a non-existing word gives error
"invalid definition count: no", when we really
want "552 no match".
changeset: 5605:33abb649cb63
user: Robert Griesemer <gri@golang.org>
date: Thu Jun 03 16:55:50 2010 -0700
files: doc/go_spec.html
description:
go spec: Base comparison compatibility on assignment compatibility.
Specifically:
- Simplified definition of comparison compatibility and folded into
section on comparison operators since it's only used there.
This is a small language change/cleanup. As a consequence:
- An interface value may now be compared against a non-interface value.
- Channels with opposite directions cannot be compared directly anymore
(per discussion with rsc).
Ian Lance Taylor [Sun, 12 Sep 2010 05:51:05 +0000 (22:51 -0700)]
test: Make gccgo believe that the variables can change.
This test is supposed to test runtime behaviour. However,
gccgo is smart enough to see that the variables are local to
the package and that they never change. Add a function so
that gccgo thinks that the variables can in fact vary, so that
the test works at runtime as expected.
If you look at the sequence of values returned
by waitpid, it simply tells us about the child of
clone before it tells us that the parent called clone.
There's nothing we can do but assume unexpected
tids are newly cloned children.
The panic NaN was a translation error.
The earliest version said panic "return sys.NaN()",
and when sys.NaN came along, it changed
to "panic sys.NaN()" instead of "return sys.NaN()".
Rob Pike [Sat, 11 Sep 2010 03:55:29 +0000 (20:55 -0700)]
arm: work around reg allocator bug in 5g, in two parts.
1) hack regalloc to leave R9 (m) and R10 (g) alone.
the real fix is tricker, but this gets us running
2) fix up the few places in the package sources that
the shortage of registers affects, by simplifying
some expressions.
all of this should be reverted when the right fix is in.
Ian Lance Taylor [Sat, 11 Sep 2010 02:12:10 +0000 (19:12 -0700)]
test: Match gccgo error messages.
The gccgo compiler does not always generate a "cannot use"
error message. It only does so for named types. Maybe that
should change, but in any case it is irrelevant for this test.
assign1.go:105:4: error: incompatible types in assignment (cannot use type A1 as type A)
assign1.go:107:5: error: incompatible types in assignment (cannot use type A as type A1)
assign1.go:112:4: error: incompatible types in assignment (cannot use type B1 as type B)
assign1.go:114:5: error: incompatible types in assignment (cannot use type B as type B1)
assign1.go:119:4: error: incompatible types in assignment (cannot use type C1 as type C)
assign1.go:121:5: error: incompatible types in assignment (cannot use type C as type C1)
assign1.go:126:4: error: incompatible types in assignment (cannot use type F1 as type F)
assign1.go:128:5: error: incompatible types in assignment (cannot use type F as type F1)
assign1.go:140:4: error: incompatible types in assignment (cannot use type M1 as type M)
assign1.go:142:5: error: incompatible types in assignment (cannot use type M as type M1)
assign1.go:147:4: error: incompatible types in assignment (cannot use type P1 as type P)
assign1.go:149:5: error: incompatible types in assignment (cannot use type P as type P1)
assign1.go:154:4: error: incompatible types in assignment (cannot use type S1 as type S)
assign1.go:156:5: error: incompatible types in assignment (cannot use type S as type S1)
assign1.go:158:6: error: incompatible types in assignment
assign1.go:159:6: error: incompatible types in assignment
assign1.go:160:5: error: incompatible types in assignment
assign1.go:161:5: error: incompatible types in assignment
assign1.go:162:6: error: incompatible types in assignment
assign1.go:163:6: error: incompatible types in assignment
assign1.go:165:6: error: incompatible types in assignment
assign1.go:166:6: error: incompatible types in assignment
assign1.go:167:5: error: incompatible types in assignment
assign1.go:168:5: error: incompatible types in assignment
assign1.go:169:6: error: incompatible types in assignment
assign1.go:170:6: error: incompatible types in assignment
assign1.go:172:6: error: incompatible types in assignment
assign1.go:173:6: error: incompatible types in assignment
assign1.go:174:5: error: incompatible types in assignment
assign1.go:175:5: error: incompatible types in assignment
assign1.go:176:6: error: incompatible types in assignment
assign1.go:177:6: error: incompatible types in assignment
assign1.go:179:6: error: incompatible types in assignment
assign1.go:180:6: error: incompatible types in assignment
assign1.go:181:5: error: incompatible types in assignment
assign1.go:182:5: error: incompatible types in assignment
assign1.go:183:6: error: incompatible types in assignment
assign1.go:184:6: error: incompatible types in assignment
assign1.go:186:6: error: incompatible types in assignment
assign1.go:187:6: error: incompatible types in assignment
assign1.go:188:5: error: incompatible types in assignment
assign1.go:189:5: error: incompatible types in assignment
assign1.go:190:6: error: incompatible types in assignment
assign1.go:191:6: error: incompatible types in assignment
assign1.go:193:6: error: incompatible types in assignment
assign1.go:194:6: error: incompatible types in assignment
assign1.go:195:5: error: incompatible types in assignment
assign1.go:196:5: error: incompatible types in assignment
assign1.go:197:6: error: incompatible types in assignment
assign1.go:198:6: error: incompatible types in assignment
assign1.go:200:6: error: incompatible types in assignment
assign1.go:201:6: error: incompatible types in assignment
assign1.go:202:5: error: incompatible types in assignment
assign1.go:203:5: error: incompatible types in assignment
assign1.go:204:6: error: incompatible types in assignment
assign1.go:205:6: error: incompatible types in assignment
assign1.go:207:6: error: incompatible types in assignment
assign1.go:208:6: error: incompatible types in assignment
assign1.go:209:5: error: incompatible types in assignment
assign1.go:210:5: error: incompatible types in assignment
assign1.go:211:6: error: incompatible types in assignment
assign1.go:212:6: error: incompatible types in assignment
Ian Lance Taylor [Fri, 10 Sep 2010 22:54:16 +0000 (15:54 -0700)]
test: Run garbage collector before testing malloc numbers.
The gccgo library generates some garbage in the init routines
because it handles interfaces slightly differently. Since the
test sets MemStats.Alloc to 0, the first time the garbage
collector runs it goes negative and the test fails.
Ian Lance Taylor [Fri, 10 Sep 2010 22:37:20 +0000 (15:37 -0700)]
test: Reduce race conditions in chan/nonblock.go.
nonblock.go wants to test nonblocking operations on
synchronous channels, so it is inherently racy. This
introduces loops to make the race conditions much more likely
to succeed when using gccgo.
Ian Lance Taylor [Fri, 10 Sep 2010 19:44:37 +0000 (12:44 -0700)]
test: Match gccgo error messages.
undef.go:12:6: error: reference to undefined name ‘x’
undef.go:13:6: error: reference to undefined name ‘x’
undef.go:14:6: error: reference to undefined name ‘x’
undef.go:22:25: error: reference to undefined name ‘y’
undef.go:42:11: error: reference to undefined name ‘v’
Ian Lance Taylor [Fri, 10 Sep 2010 05:40:25 +0000 (22:40 -0700)]
test: Match gccgo error messages.
With the recursive descent parser that gccgo uses, I think
that it doesn't make sense to try to match a statement where a
statement is not expected. If the construct is not a
statement, you will just get bizarre error messages.
Rob Pike [Thu, 9 Sep 2010 04:28:56 +0000 (14:28 +1000)]
runtime: fix bug in tracebacks
Need to correct for deferproc's extra 2 words of stack or in some
cases (such as memory profiling) traceback can cause a crash.
Also bulletproof the closure test.