On Wed, 31 Oct 2007, Bruce Evans wrote:
__mb_sb_limit is extern int, so the ABI breakage was obvious. If it
had been a compile-time constant with the usual ${CFLAGS} but not a
constant with -O0, of if the constant depended on ${CFLAGS} or changed
with __FreeBSD_version, then the problem would h
On Wed, Oct 31, 2007 at 09:41:13PM +1100, Bruce Evans wrote:
> Now the limit is spelled 128, so it really is constant and won't cause
> problems unless the constant changes.
Only in one particular case, the rest cases still need __mb_sb_limit.
> _ctype.h still declares __mb_sb_limit but now doesn
On Thu, 25 Oct 2007, LI Xin wrote:
Scott Long wrote:
Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 09:15:27AM -0600, Scott Long wrote:
Andrey A. Chernov wrote:
ache2007-10-15 09:51:30 UTC
FreeBSD src repository
Modified files:
lib/libc/locale utf8.c Log:
Add comment
Guys,
Just FYI, rafan@ has posted a latest version of patch, which will
hopefully fix the ABI compatibility issue, while retaining ache@'s fixes
to utf-8 handling. The drawback is that we will lose some inline macros
which could (presumably) give a performance penalty for heavy ctype(3)
users, bu
David O'Brien wrote:
On Sat, Oct 27, 2007 at 11:57:05PM +0200, Kris Kennaway wrote:
David O'Brien wrote:
On Fri, Oct 26, 2007 at 05:46:09PM -0400, Ken Smith wrote:
On Fri, 2007-10-26 at 13:30 -0700, David O'Brien wrote:
On Fri, Oct 26, 2007 at 02:47:13PM -0400, Ken Smith wrote:
If portmgr@ u
On Sat, Oct 27, 2007 at 11:57:05PM +0200, Kris Kennaway wrote:
> David O'Brien wrote:
>> On Fri, Oct 26, 2007 at 05:46:09PM -0400, Ken Smith wrote:
>>> On Fri, 2007-10-26 at 13:30 -0700, David O'Brien wrote:
On Fri, Oct 26, 2007 at 02:47:13PM -0400, Ken Smith wrote:
> If portmgr@ uses RELE
David O'Brien wrote:
On Fri, Oct 26, 2007 at 05:46:09PM -0400, Ken Smith wrote:
On Fri, 2007-10-26 at 13:30 -0700, David O'Brien wrote:
On Fri, Oct 26, 2007 at 02:47:13PM -0400, Ken Smith wrote:
If portmgr@ uses RELENG_6 to base their builds on breaking forward
compat would mean all consumers
On Fri, Oct 26, 2007 at 05:46:09PM -0400, Ken Smith wrote:
> On Fri, 2007-10-26 at 13:30 -0700, David O'Brien wrote:
> > On Fri, Oct 26, 2007 at 02:47:13PM -0400, Ken Smith wrote:
> > > If portmgr@ uses RELENG_6 to base their builds on breaking forward
> > > compat would mean all consumers of packa
John Baldwin wrote:
On Friday 26 October 2007 12:00:54 pm Ken Smith wrote:
On Fri, 2007-10-26 at 11:41 -0400, John Baldwin wrote:
On Friday 26 October 2007 10:53:47 am David O'Brien wrote:
On Thu, Oct 25, 2007 at 05:31:03PM -0400, Ken Smith wrote:
What we need to try and avoid unless *absolut
On Fri, Oct 26, 2007 at 04:26:41PM -0700, John-Mark Gurney wrote:
> > Although standard permits functions only, I object against permanent
> > removing inlines from ctype. Almost every system have them as macros or
> > inlines and not as function calls, and it is for reason. Ctype functions
> >
Andrey Chernov wrote this message on Fri, Oct 26, 2007 at 23:00 +0400:
> On Sat, Oct 27, 2007 at 12:54:21AM +0800, Rong-en Fan wrote:
> > As for RELENG_7 and HEAD, I'm not 100% sure whether we should restore
> > the compatibility as 1) 7.0 is not released yet, 2) we don't promise
> > anything in
>
On Fri, 2007-10-26 at 13:30 -0700, David O'Brien wrote:
> On Fri, Oct 26, 2007 at 02:47:13PM -0400, Ken Smith wrote:
> > If portmgr@ uses RELENG_6 to base their builds on breaking forward
> > compat would mean all consumers of packages-6-stable would also need
> > to use RELENG_6.
>
> When did thi
On Fri, Oct 26, 2007 at 01:32:52PM -0700, LI Xin wrote:
> David O'Brien wrote:
> > On Fri, Oct 26, 2007 at 02:47:13PM -0400, Ken Smith wrote:
> >> If portmgr@ uses RELENG_6 to base their builds on breaking forward
> >> compat would mean all consumers of packages-6-stable would also need
> >> to use
David O'Brien wrote:
> On Fri, Oct 26, 2007 at 02:47:13PM -0400, Ken Smith wrote:
>> If portmgr@ uses RELENG_6 to base their builds on breaking forward
>> compat would mean all consumers of packages-6-stable would also need
>> to use RELENG_6.
>
> When did this sease being a requirement?
> 'pkg_ad
On Fri, Oct 26, 2007 at 02:47:13PM -0400, Ken Smith wrote:
> If portmgr@ uses RELENG_6 to base their builds on breaking forward
> compat would mean all consumers of packages-6-stable would also need
> to use RELENG_6.
When did this sease being a requirement?
'pkg_add -r' certainly enforces this re
Andrey Chernov wrote:
On Sat, Oct 27, 2007 at 12:54:21AM +0800, Rong-en Fan wrote:
As for RELENG_7 and HEAD, I'm not 100% sure whether we should restore
the compatibility as 1) 7.0 is not released yet, 2) we don't promise
anything in
HEAD branch. However, if most people think get rid of inline s
On Fri, Oct 26, 2007 at 10:41:14AM -0700, LI Xin wrote:
> rafan@ has a patch that implements Warner's suggestion by removing the
> inline's from _ctype.h and the macros from ctype.h, and is testing the
> impact. Basically the approach do the following:
>
> - Keep __mb_sb_limit intact. It is use
On Sat, Oct 27, 2007 at 12:54:21AM +0800, Rong-en Fan wrote:
> As for RELENG_7 and HEAD, I'm not 100% sure whether we should restore
> the compatibility as 1) 7.0 is not released yet, 2) we don't promise
> anything in
> HEAD branch. However, if most people think get rid of inline stuffs and
> have
On Fri, 2007-10-26 at 12:24 -0400, John Baldwin wrote:
> I was more concerned about there being a
> new general policy. Are you really sure you want forwards compat and not
> just backwards compat ABI?
I certainly don't want to guarantee it. But I'd like to avoid breaking
forwards compat if the
On Thu, Oct 25, 2007 at 06:12:16PM -0600, M. Warner Losh wrote:
> Maybe we should make them *NOT* be inline instead of all these
> kludges?
In case we trade compatibility for slowdown in some apps, yes. They are
inlines for reason, almost every system have them as inlines.
--
http://ache.pp.ru/
On Friday 26 October 2007 12:00:54 pm Ken Smith wrote:
> On Fri, 2007-10-26 at 11:41 -0400, John Baldwin wrote:
> > On Friday 26 October 2007 10:53:47 am David O'Brien wrote:
> > > On Thu, Oct 25, 2007 at 05:31:03PM -0400, Ken Smith wrote:
> > > > What we need to try and avoid unless *absolutely* *
Scott Long wrote:
> You're right that nothing can be promised, but you're ignoring what Ken
> and I are saying about having a justification for changes. Yes, MSI
> added a feature that wasn't present in previous releases. But it's
> something that is optional and can be easily ifdef'd in source.
Ken Smith wrote:
> On Fri, 2007-10-26 at 11:41 -0400, John Baldwin wrote:
>> On Friday 26 October 2007 10:53:47 am David O'Brien wrote:
>>> On Thu, Oct 25, 2007 at 05:31:03PM -0400, Ken Smith wrote:
What we need to try and avoid unless *absolutely* *necessary* is the
[...]
using a 6.3-REL
On Fri, Oct 26, 2007 at 10:37:22AM -0600, Scott Long wrote:
> and I are saying about having a justification for changes.
..
> The ctype ABI change
> isn't like that at all; it's mandatory and it can't be
> worked around. We also asked that even justifiable changes be evaluated
> to see if there w
On 10/27/07, Scott Long <[EMAIL PROTECTED]> wrote:
> John Baldwin wrote:
> > On Thursday 25 October 2007 07:04:08 pm Scott Long wrote:
> >> Robert Watson wrote:
> >>> On Thu, 25 Oct 2007, Andrey Chernov wrote:
> >>>
> On Thu, Oct 25, 2007 at 12:05:40PM -0700, LI Xin wrote:
> > Well, I thin
John Baldwin wrote:
On Thursday 25 October 2007 07:04:08 pm Scott Long wrote:
Robert Watson wrote:
On Thu, 25 Oct 2007, Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 12:05:40PM -0700, LI Xin wrote:
Well, I think the problem is not exposing a new symbol by itself, but
__mb_sb_limit is being u
On Fri, Oct 26, 2007 at 11:44:33AM -0400, John Baldwin wrote:
> On Thursday 25 October 2007 07:04:08 pm Scott Long wrote:
> > Robert Watson wrote:
> > > On Thu, 25 Oct 2007, Andrey Chernov wrote:
> > >
> > >> On Thu, Oct 25, 2007 at 12:05:40PM -0700, LI Xin wrote:
> > >>> Well, I think the problem
On Fri, 2007-10-26 at 11:41 -0400, John Baldwin wrote:
> On Friday 26 October 2007 10:53:47 am David O'Brien wrote:
> > On Thu, Oct 25, 2007 at 05:31:03PM -0400, Ken Smith wrote:
> > > What we need to try and avoid unless *absolutely* *necessary* is the
> > > part Scott quoted above - binaries comp
On Fri, 26 Oct 2007, John Baldwin wrote:
I think the issue here is that the change occurred very quickly after the
branch, and when users wanted to 'change gears' back to RELENG_7 from HEAD
once it was created immediately ran into the problem. It seems like a
useful piece of post-branch advic
On Thursday 25 October 2007 07:04:08 pm Scott Long wrote:
> Robert Watson wrote:
> > On Thu, 25 Oct 2007, Andrey Chernov wrote:
> >
> >> On Thu, Oct 25, 2007 at 12:05:40PM -0700, LI Xin wrote:
> >>> Well, I think the problem is not exposing a new symbol by itself, but
> >>> __mb_sb_limit is being
On Friday 26 October 2007 10:53:47 am David O'Brien wrote:
> On Thu, Oct 25, 2007 at 05:31:03PM -0400, Ken Smith wrote:
> > What we need to try and avoid unless *absolutely* *necessary* is the
> > part Scott quoted above - binaries compiled on 6.3-REL should work on
> > 6.2-REL unless there was a r
On Thu, Oct 25, 2007 at 05:31:03PM -0400, Ken Smith wrote:
> What we need to try and avoid unless *absolutely* *necessary* is the
> part Scott quoted above - binaries compiled on 6.3-REL should work on
> 6.2-REL unless there was a really big issue and the solution to that
> issue required us to bre
On 10/26/07, Ken Smith <[EMAIL PROTECTED]> wrote:
> On Thu, 2007-10-25 at 23:14 +0400, Andrey Chernov wrote:
> > On Thu, Oct 25, 2007 at 01:10:13PM -0600, Scott Long wrote:
> > >> Well, I think the problem is not exposing a new symbol by itself, but
> > >> __mb_sb_limit is being used in _ctype.h, i
In message: <[EMAIL PROTECTED]>
LI Xin <[EMAIL PROTECTED]> writes:
: Andrey Chernov wrote:
: > On Thu, Oct 25, 2007 at 12:30:07PM -0600, Scott Long wrote:
: [snip]
: >> and is causing many users problems
: >> during upgrades and normal operations.
: >
: > Like what? I don't observe a
Robert Watson wrote:
On Thu, 25 Oct 2007, Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 12:05:40PM -0700, LI Xin wrote:
Well, I think the problem is not exposing a new symbol by itself, but
__mb_sb_limit is being used in _ctype.h, in a form of __inline
functions. Therefore, the change will br
On Thu, 25 Oct 2007, Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 12:05:40PM -0700, LI Xin wrote:
Well, I think the problem is not exposing a new symbol by itself, but
__mb_sb_limit is being used in _ctype.h, in a form of __inline functions.
Therefore, the change will break new binaries runni
On Thu, 2007-10-25 at 23:14 +0400, Andrey Chernov wrote:
> On Thu, Oct 25, 2007 at 01:10:13PM -0600, Scott Long wrote:
> >> Well, I think the problem is not exposing a new symbol by itself, but
> >> __mb_sb_limit is being used in _ctype.h, in a form of __inline
> >> functions. Therefore, the chang
On Thu, Oct 25, 2007 at 11:57:09PM +0400, Andrey Chernov wrote:
> On Thu, Oct 25, 2007 at 03:31:20PM -0400, Daniel Eischen wrote:
> >> Although standard permit this, it will cause lots of programs to break
> >> here and there and even may cause core dump. I don't think it is what
> >> everybody wan
On Thu, Oct 25, 2007 at 03:17:32PM -0400, Daniel Eischen wrote:
> Why not simply avoid the error check for inlines. Solaris
> doesn't seem to add a check, it just does the conversion.
> The non-inlined versions do error checking.
I already answer that several times. Nobody wants 80% programs ethe
On Thu, Oct 25, 2007 at 03:31:20PM -0400, Daniel Eischen wrote:
>> Although standard permit this, it will cause lots of programs to break
>> here and there and even may cause core dump. I don't think it is what
>> everybody wants. Non-inlined versions a) not used nowdays b) must be exact
>> as inli
On Thu, 25 Oct 2007, Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 12:20:16PM -0700, LI Xin wrote:
Daniel Eischen wrote:
Why not simply avoid the error check for inlines. Solaris
doesn't seem to add a check, it just does the conversion.
The non-inlined versions do error checking.
I'll take
On Thu, 25 Oct 2007, Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 03:17:32PM -0400, Daniel Eischen wrote:
Why not simply avoid the error check for inlines. Solaris
doesn't seem to add a check, it just does the conversion.
The non-inlined versions do error checking.
Although standard permit
On Thu, Oct 25, 2007 at 12:20:16PM -0700, LI Xin wrote:
> Daniel Eischen wrote:
> > Why not simply avoid the error check for inlines. Solaris
> > doesn't seem to add a check, it just does the conversion.
> > The non-inlined versions do error checking.
>
> I'll take a look at this and provide a pa
On Thu, Oct 25, 2007 at 03:17:32PM -0400, Daniel Eischen wrote:
> Why not simply avoid the error check for inlines. Solaris
> doesn't seem to add a check, it just does the conversion.
> The non-inlined versions do error checking.
Although standard permit this, it will cause lots of programs to br
On Thu, Oct 25, 2007 at 12:15:13PM -0700, LI Xin wrote:
> Andrey Chernov wrote:
> > On Thu, Oct 25, 2007 at 12:30:07PM -0600, Scott Long wrote:
> [snip]
> >> and is causing many users problems
> >> during upgrades and normal operations.
> >
> > Like what? I don't observe any singe problem so far
Daniel Eischen wrote:
> Why not simply avoid the error check for inlines. Solaris
> doesn't seem to add a check, it just does the conversion.
> The non-inlined versions do error checking.
I'll take a look at this and provide a patch for review.
Cheers,
--
Xin LI <[EMAIL PROTECTED]> http://
On Thu, 25 Oct 2007, LI Xin wrote:
Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 09:44:06AM -0700, LI Xin wrote:
Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 09:15:27AM -0600, Scott Long wrote:
Andrey A. Chernov wrote:
ache2007-10-15 09:51:30 UTC
FreeBSD src repository
Modified
Andrey Chernov wrote:
> On Thu, Oct 25, 2007 at 12:30:07PM -0600, Scott Long wrote:
[snip]
>> and is causing many users problems
>> during upgrades and normal operations.
>
> Like what? I don't observe any singe problem so far.
This has revealed another issue on RELENG_6 while rafan@ did the MF
On Thu, Oct 25, 2007 at 01:10:13PM -0600, Scott Long wrote:
>> Well, I think the problem is not exposing a new symbol by itself, but
>> __mb_sb_limit is being used in _ctype.h, in a form of __inline
>> functions. Therefore, the change will break new binaries running on
>> older systems. Personall
On Thu, Oct 25, 2007 at 12:05:40PM -0700, LI Xin wrote:
> Well, I think the problem is not exposing a new symbol by itself, but
> __mb_sb_limit is being used in _ctype.h, in a form of __inline
> functions. Therefore, the change will break new binaries running on
> older systems.
Yes. Only vice
LI Xin wrote:
Scott Long wrote:
Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 09:15:27AM -0600, Scott Long wrote:
Andrey A. Chernov wrote:
ache2007-10-15 09:51:30 UTC
FreeBSD src repository
Modified files:
lib/libc/locale utf8.c Log:
Add comment explaining __mb_sb_lim
Andrey Chernov wrote:
> On Thu, Oct 25, 2007 at 09:44:06AM -0700, LI Xin wrote:
>> Andrey Chernov wrote:
>>> On Thu, Oct 25, 2007 at 09:15:27AM -0600, Scott Long wrote:
Andrey A. Chernov wrote:
> ache2007-10-15 09:51:30 UTC
> FreeBSD src repository
> Modified files:
>>>
Scott Long wrote:
> Andrey Chernov wrote:
>> On Thu, Oct 25, 2007 at 09:15:27AM -0600, Scott Long wrote:
>>> Andrey A. Chernov wrote:
ache2007-10-15 09:51:30 UTC
FreeBSD src repository
Modified files:
lib/libc/locale utf8.c Log:
Add comment explaini
On Thu, Oct 25, 2007 at 12:30:07PM -0600, Scott Long wrote:
> Exposing the __mb_sb_limit symbol has instantly created a need
> to have a compat7x package,
Yes. We supposed to do this in any case.
> and is causing many users problems
> during upgrades and normal operations.
Like what? I don't
On Thu, Oct 25, 2007 at 09:44:06AM -0700, LI Xin wrote:
> Andrey Chernov wrote:
> > On Thu, Oct 25, 2007 at 09:15:27AM -0600, Scott Long wrote:
> >> Andrey A. Chernov wrote:
> >>> ache2007-10-15 09:51:30 UTC
> >>> FreeBSD src repository
> >>> Modified files:
> >>> lib/libc/locale
Andrey Chernov wrote:
On Thu, Oct 25, 2007 at 09:15:27AM -0600, Scott Long wrote:
Andrey A. Chernov wrote:
ache2007-10-15 09:51:30 UTC
FreeBSD src repository
Modified files:
lib/libc/locale utf8.c Log:
Add comment explaining __mb_sb_limit trick here.
Revision Chang
Andrey Chernov wrote:
> On Thu, Oct 25, 2007 at 09:15:27AM -0600, Scott Long wrote:
>> Andrey A. Chernov wrote:
>>> ache2007-10-15 09:51:30 UTC
>>> FreeBSD src repository
>>> Modified files:
>>> lib/libc/locale utf8.c Log:
>>> Add comment explaining __mb_sb_limit trick here
On Thu, Oct 25, 2007 at 09:15:27AM -0600, Scott Long wrote:
> Andrey A. Chernov wrote:
>> ache2007-10-15 09:51:30 UTC
>> FreeBSD src repository
>> Modified files:
>> lib/libc/locale utf8.c Log:
>> Add comment explaining __mb_sb_limit trick here.
>> Revision Changes
Andrey A. Chernov wrote:
ache2007-10-15 09:51:30 UTC
FreeBSD src repository
Modified files:
lib/libc/locale utf8.c
Log:
Add comment explaining __mb_sb_limit trick here.
Revision ChangesPath
1.16 +5 -0 src/lib/libc/locale/utf8.c
When is the ABI
59 matches
Mail list logo
