Discussion:
[ast-developers] AT&T Software Technology ast alpha software download update
Glenn Fowler
2013-07-27 16:17:45 UTC
Permalink
the AT&T Software Technology ast alpha 2013-07-27 source release
has been posted to the download site
http://www.research.att.com/sw/download/alpha/
the package names and md5 checksums are
INIT 8cdf2460c146a7412ee1f2d37d19e3ce
ast-base 46a03d0e49840acedef8c0623cbef85c
ast-open f61766d8cb1f91c960a3a2a1385adcd6
ast-ksh 911d28c7263e90ec1e8037ad0b6db643
the md5 sums should match the ones listed on the download page

still verifying builds on some architectures
should be ok on linux & solaris
for the next week or so stability/regression patches will take priority

we have been seeing spurious "make: lost connection with coshell" in ast-open builds
the current theory is that a signal is leaking through unchecked
it may be a flaw in the signal restart mechanism or a syscall that we missed
and requires an EINTR restart loop -- if anyone catches this one in the act
it would be a big help

changes since 2013-07-19

:::::::: ksh93 ::::::::

13-07-24 --- Release ksh93v- ---
13-07-24 A bug in which += assignments did not work for environment variables
specified for non-special builtin commands has been fixed.
13-07-24 The _ variable is new set as a reference to the type inside
discipline functions for non-type variables in the type.
13-07-23 A bug in the arithmetic assignment to an enum using a ?: operator
with the nounset option has been fixed.
13-07-22 The .sh.sig variable has been modified to treat .sh.value as a
compound variable containing int and ptr.

:::::::: libast ::::::::

13-07-25 add ast/sys/socket.h => ast_socket.h => ../include/sys/socket.h for intercepts
13-07-22 pathopen.c: fix PATH_DEV logic to preserve ksh sh_open() semantics
13-07-19 fix features/fcntl.c to keep track of highest bit in o_local

:::::::: libcoshell ::::::::

13-07-24 coopen.c: close child side of msgfp in parent (stops hung caller on shell failure)
Lionel Cons
2013-07-27 16:32:15 UTC
Permalink
Post by Glenn Fowler
the AT&T Software Technology ast alpha 2013-07-27 source release
has been posted to the download site
http://www.research.att.com/sw/download/alpha/
the package names and md5 checksums are
INIT 8cdf2460c146a7412ee1f2d37d19e3ce
ast-base 46a03d0e49840acedef8c0623cbef85c
ast-open f61766d8cb1f91c960a3a2a1385adcd6
ast-ksh 911d28c7263e90ec1e8037ad0b6db643
the md5 sums should match the ones listed on the download page
still verifying builds on some architectures
should be ok on linux & solaris
for the next week or so stability/regression patches will take priority
we have been seeing spurious "make: lost connection with coshell" in ast-open builds
the current theory is that a signal is leaking through unchecked
it may be a flaw in the signal restart mechanism or a syscall that we missed
and requires an EINTR restart loop -- if anyone catches this one in the act
it would be a big help
There have been such coshell warnings in the ast-open builds since a
year or longer. Are you sure this one is different?

Lionel
Glenn Fowler
2013-07-28 04:42:53 UTC
Permalink
Post by Lionel Cons
Post by Glenn Fowler
the AT&T Software Technology ast alpha 2013-07-27 source release
has been posted to the download site
http://www.research.att.com/sw/download/alpha/
the package names and md5 checksums are
INIT 8cdf2460c146a7412ee1f2d37d19e3ce
ast-base 46a03d0e49840acedef8c0623cbef85c
ast-open f61766d8cb1f91c960a3a2a1385adcd6
ast-ksh 911d28c7263e90ec1e8037ad0b6db643
the md5 sums should match the ones listed on the download page
still verifying builds on some architectures
should be ok on linux & solaris
for the next week or so stability/regression patches will take priority
we have been seeing spurious "make: lost connection with coshell" in ast-open builds
the current theory is that a signal is leaking through unchecked
it may be a flaw in the signal restart mechanism or a syscall that we missed
and requires an EINTR restart loop -- if anyone catches this one in the act
it would be a big help
There have been such coshell warnings in the ast-open builds since a
year or longer. Are you sure this one is different?
not sure
Irek Szczesniak
2013-07-27 18:26:16 UTC
Permalink
Post by Glenn Fowler
the AT&T Software Technology ast alpha 2013-07-27 source release
has been posted to the download site
http://www.research.att.com/sw/download/alpha/
the package names and md5 checksums are
INIT 8cdf2460c146a7412ee1f2d37d19e3ce
ast-base 46a03d0e49840acedef8c0623cbef85c
ast-open f61766d8cb1f91c960a3a2a1385adcd6
ast-ksh 911d28c7263e90ec1e8037ad0b6db643
the md5 sums should match the ones listed on the download page
still verifying builds on some architectures
should be ok on linux & solaris
for the next week or so stability/regression patches will take priority
we have been seeing spurious "make: lost connection with coshell" in ast-open builds
the current theory is that a signal is leaking through unchecked
it may be a flaw in the signal restart mechanism or a syscall that we missed
and requires an EINTR restart loop -- if anyone catches this one in the act
it would be a big help
changes since 2013-07-19
13-07-24 --- Release ksh93v- ---
13-07-24 A bug in which += assignments did not work for environment variables
specified for non-special builtin commands has been fixed.
13-07-24 The _ variable is new set as a reference to the type inside
discipline functions for non-type variables in the type.
13-07-23 A bug in the arithmetic assignment to an enum using a ?: operator
with the nounset option has been fixed.
13-07-22 The .sh.sig variable has been modified to treat .sh.value as a
compound variable containing int and ptr.
XX-XX-XX A bug was fixed which caused ksh93 to close shp->pwd
prematurely for the last command executed by the shell
(close-on-exec).
I refrain from using more explicit language to describe what this bug did to us.

I'd also like to request to change this entry:
13-05-22 +Added -c option to typeset which is similar to typeset -m except
that the variable is copied rather than moved. This is experimental
and there are known bugs. If it can't be made reliable it will
not be added.
and remove the 'This is experimental.*it will not be added.' part.
IMO typeset -c has proven its usefulness. The remainder is just fixing
all damn bugs.

Irek
ольга крыжановская
2013-07-29 09:16:35 UTC
Permalink
Solaris 11/i386, 64bit build has a few problems:

1. The sigchld test fails in compiled shell script mode:
test sigchld begins at 2013-07-29+10:26:31
test sigchld passed at 2013-07-29+10:26:59 [ 19 tests 0 errors ]
test sigchld(shcomp) begins at 2013-07-29+10:27:00
[193]: 29019 already finished and reaped
childsig
[193]: 29019 already finished and reaped
[193]: 29019 already finished and reaped
childsig
test sigchld(shcomp) failed at 2013-07-29+10:27:28 with exit code 3 [
19 tests 3 errors ]

2. signals.sh fails:
test signal begins at 2013-07-29+10:27:28
./src/cmd/ksh93/tests/shtests[333]: eval: line 1: 29076: Terminated
test signal failed at 2013-07-29+10:28:56 with exit code 271 [ 47
tests 271 errors ]
test signal(shcomp) begins at 2013-07-29+10:28:56
./src/cmd/ksh93/tests/shtests: line 354: 29763: Terminated
test signal(shcomp) failed at 2013-07-29+10:30:23 with exit code 271 [
47 tests 271 errors ]

Olga
Post by Glenn Fowler
the AT&T Software Technology ast alpha 2013-07-27 source release
has been posted to the download site
http://www.research.att.com/sw/download/alpha/
the package names and md5 checksums are
INIT 8cdf2460c146a7412ee1f2d37d19e3ce
ast-base 46a03d0e49840acedef8c0623cbef85c
ast-open f61766d8cb1f91c960a3a2a1385adcd6
ast-ksh 911d28c7263e90ec1e8037ad0b6db643
the md5 sums should match the ones listed on the download page
still verifying builds on some architectures
should be ok on linux & solaris
for the next week or so stability/regression patches will take priority
we have been seeing spurious "make: lost connection with coshell" in ast-open builds
the current theory is that a signal is leaking through unchecked
it may be a flaw in the signal restart mechanism or a syscall that we missed
and requires an EINTR restart loop -- if anyone catches this one in the act
it would be a big help
changes since 2013-07-19
13-07-24 --- Release ksh93v- ---
13-07-24 A bug in which += assignments did not work for environment variables
specified for non-special builtin commands has been fixed.
13-07-24 The _ variable is new set as a reference to the type inside
discipline functions for non-type variables in the type.
13-07-23 A bug in the arithmetic assignment to an enum using a ?: operator
with the nounset option has been fixed.
13-07-22 The .sh.sig variable has been modified to treat .sh.value as a
compound variable containing int and ptr.
13-07-25 add ast/sys/socket.h => ast_socket.h => ../include/sys/socket.h for intercepts
13-07-22 pathopen.c: fix PATH_DEV logic to preserve ksh sh_open() semantics
13-07-19 fix features/fcntl.c to keep track of highest bit in o_local
13-07-24 coopen.c: close child side of msgfp in parent (stops hung caller on shell failure)
_______________________________________________
ast-developers mailing list
ast-developers at lists.research.att.com
http://lists.research.att.com/mailman/listinfo/ast-developers
--
, _ _ ,
{ \/`o;====- Olga Kryzhanovska -====;o`\/ }
.----'-/`-/ olga.kryzhanovska at gmail.com \-`\-'----.
`'-..-| / http://twitter.com/fleyta \ |-..-'`
/\/\ Solaris/BSD//C/C++ programmer /\/\
`--` `--`
David Korn
2013-07-29 14:12:14 UTC
Permalink
Post by ольга крыжановская
test sigchld begins at 2013-07-29+10:26:31
test sigchld passed at 2013-07-29+10:26:59 [ 19 tests 0 errors ]
test sigchld(shcomp) begins at 2013-07-29+10:27:00
[193]: 29019 already finished and reaped
childsig
[193]: 29019 already finished and reaped
[193]: 29019 already finished and reaped
childsig
test sigchld(shcomp) failed at 2013-07-29+10:27:28 with exit code 3 [
19 tests 3 errors ]
This is a known bug on all systems and is low priority since it is unlikely
to occur in practice. It is on my bug list.

David Korn
dgk at research.att.com

Loading...