Previously Marcus Brinkmann wrote:
As for me personally, I have made peace with -O2 code. stepi is your friend
;)
Fwiw, debugging C++ code using STL with -O2 is almost impossible.
Wichert.
--
_
/[EMAIL PROTECTED]
[ Sorry if this comes through twice, my Postfix/TLS certificate expired,
and mail stopped working for a bit ]
Here's an updated patch which should address the concerns raised.
--- debian-policy-3.5.6.1/policy.sgml 2002-03-14 13:17:48.0 -0500
+++ debian-policy-3.5.6.1.hacked/policy.sgml
On 21 Aug 2002 17:30:30 -0400,
Colin Walters [EMAIL PROTECTED] wrote:
ifeq (,$(findstring nostrip,$(DEB_BUILD_OPTIONS)))
INSTALL_PROGRAM += -s
endif
I saw a library which did install -s *.a (already fixed).
--
Oohara Yuuma [EMAIL PROTECTED]
Debian developer
PGP key (key ID F464A695)
Marcus Brinkmann [EMAIL PROTECTED] writes:
You might want to add a warning that this needs to be tested. Some
packages, like glibc or the Hurd, can not be built without optimization
(for example because of inline functions not being inlined).
It should be pointed out that this fact
On Mon, 2002-08-19 at 20:35, Richard Braakman wrote:
Note that many packages don't currently do this. I often had to do
horrible things to Makefiles to get a binary with debugging symbols.
I speak in past tense because I usually don't bother to do this anymore.
So listen up, maintainers: if
On 18 Aug 2002 20:18:43 -0400,
Colin Walters [EMAIL PROTECTED] wrote:
+ Although binaries in the build tree should be compiled with
+ debugging information by default,
How can I do it without wasting autobuilder's CPU time?
--
Oohara Yuuma [EMAIL PROTECTED]
Debian developer
PGP key
On Tue, Aug 20, 2002 at 02:00:41PM +0900, Oohara Yuuma wrote:
On 18 Aug 2002 20:18:43 -0400,
Colin Walters [EMAIL PROTECTED] wrote:
+ Although binaries in the build tree should be compiled with
+ debugging information by default,
How can I do it without wasting autobuilder's CPU
[EMAIL PROTECTED] (Oohara Yuuma) writes:
On 18 Aug 2002 20:18:43 -0400,
Colin Walters [EMAIL PROTECTED] wrote:
+ Although binaries in the build tree should be compiled with
+ debugging information by default,
How can I do it without wasting autobuilder's CPU time?
Don't worry
On 18 Aug 2002 18:16:43 -0400,
Colin Walters [EMAIL PROTECTED] wrote:
+ tagnoopt/tag
+ item
+ p
+ The presence of this string means that the package
+ should be complied with the minumum possible amount of
+ optimization. For C
On Sun, 2002-08-18 at 20:19, Oohara Yuuma wrote:
-O0 is the default of gcc. Why do I have to add it explicitly?
Well, you don't strictly speaking have to do anything yet;
DEB_BUILD_OPTIONS is a recommendation of policy, not a requirement.
Anyways, I just mentioned adding -O0 just for
On Mon, Aug 19, 2002 at 09:19:02AM +0900, Oohara Yuuma wrote:
On 18 Aug 2002 18:16:43 -0400,
Colin Walters [EMAIL PROTECTED] wrote:
+ tagnoopt/tag
+ item
+ p
+ The presence of this string means that the package
+ should be complied with the
Julian Gilbey wrote:
On Mon, Aug 19, 2002 at 09:19:02AM +0900, Oohara Yuuma wrote:
-O0 is the default of gcc. Why do I have to add it explicitly?
I don't want a bug filed just because -O0 is missing.
Good point. May in some cases be worth adding, though, if upstream
makefiles add -O2
On Sun, Aug 18, 2002 at 06:48:24PM -0500, Branden Robinson wrote:
(BTW, someone's mailer is a complete piece of crap. What the F is up
with mangling the subject line like that?)
The BTS doesn't like the [ at the start of the Subject line, for some reason.
--
2. That which causes joy or
On Mon, Aug 19, 2002 at 01:14:31PM +0200, Josip Rodin wrote:
On Sun, Aug 18, 2002 at 06:48:24PM -0500, Branden Robinson wrote:
(BTW, someone's mailer is a complete piece of crap. What the F is up
with mangling the subject line like that?)
The BTS doesn't like the [ at the start of the
At 10:59 pm, Monday, August 19 2002, Colin Walters mumbled:
I kind of suspected so, but not having access to authoritative data I
didn't want to try to change two things at once. Well, here's an
updated patch which combines both then. I also removed some old cruft
about a.out and -N.
Any
On Mon, 19 Aug 2002 13:05:18 +0200,
Ivo Timmermans [EMAIL PROTECTED] wrote:
Julian Gilbey wrote:
Good point. May in some cases be worth adding, though, if upstream
makefiles add -O2 automatically.
Same goes for -fno-omit-frame-pointer
I heard when I was in the NM queue that
On Mon, 19 Aug 2002, Steve Kowalik wrote:
At 10:59 pm, Monday, August 19 2002, Colin Walters mumbled:
I kind of suspected so, but not having access to authoritative data I
didn't want to try to change two things at once. Well, here's an
updated patch which combines both then. I also
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Colin Walters [EMAIL PROTECTED] writes:
Any seconds?
seconded.
- --
James
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Processed by Mailcrypt 3.5.6 http://mailcrypt.sourceforge.net/
[EMAIL PROTECTED] (Colin Walters) writes:
Any seconds?
I note a typo in the first line of your change:
+ By default, when a package is being built, it any binaries
Want to lose the word 'it' in that line?
Other than that, I second this proposal.
Bdale
On Mon, 2002-08-19 at 18:07, Bdale Garbee wrote:
I note a typo in the first line of your change:
+ By default, when a package is being built, it any binaries
Want to lose the word 'it' in that line?
Yep, good catch. Thanks.
On Sun, Aug 18, 2002 at 08:18:43PM -0400, Colin Walters wrote:
+ Although binaries in the build tree should be compiled with
+ debugging information by default,
Note that many packages don't currently do this. I often had to do
horrible things to Makefiles to get a binary with
[ No need to CC me, btw, despite the evil the BTS does to Reply-To ]
On Sat, 2002-08-17 at 23:14, Marcus Brinkmann wrote:
You might want to add a warning that this needs to be tested. Some
packages, like glibc or the Hurd, can not be built without optimization
(for example because of inline
This would lose a feature that I find valuable: usually, recompiling
a package with the debug option will generate a binary whose symbols
are compatible with the normal packaged binary. I have used this
several times to chase down hard-to-find library compatibility bugs,
or to interpret
On Sun, 18 Aug 2002, Richard Braakman wrote:
This would lose a feature that I find valuable: usually, recompiling
a package with the debug option will generate a binary whose symbols
are compatible with the normal packaged binary. I have used this
several times to chase down hard-to-find
Please see also my comments in #113525.
Ian.
On Sun, 2002-08-18 at 15:45, Ian Jackson wrote:
Please see also my comments in #113525.
I personally have no strong opinion on this, really. Whoever wrote the
Rationale: part of this section obviously disagrees with you, though.
Maybe machines are fast enough today in general that it isn't as
On Sun, 2002-08-18 at 05:46, Richard Braakman wrote:
For that matter, there are also Heisenbugs to consider: some bugs only
appear in the optimized version, not the un-optimized. Compiling the
debugging version with different optimization flags would make tracking
down these bugs needlessly
On Sun, Aug 18, 2002 at 06:18:45PM -0400, Colin Walters wrote:
On Sun, 2002-08-18 at 15:45, Ian Jackson wrote:
Please see also my comments in #113525.
I personally have no strong opinion on this, really. Whoever wrote the
Rationale: part of this section obviously disagrees with you,
On Sun, Aug 18, 2002 at 11:52:54PM +0100, James Troup wrote:
Yes, but it's complete and utter crap; it was then and still is now.
Ben ran a buildd on vore which is one of the faster buildds, but I've
run both vore and some of our slower (arm, m68k) buildds and I can
guarantee you that the
optimization when DEB_BUILD_OPTIONS
contains debug
Merged 113525 157131.
retitle 157131 [PROPOSAL] Rework DEB_BUILD_OPTIONS section
Bug#157131: [PROPOSAL] Suggest to minimize optimization when DEB_BUILD_OPTIONS
contains debug
Changed Bug title.
thanks
Stopping processing here.
Please contact
severity 113525 wishlist
merge 157131 113525
retitle 157131 [PROPOSAL] Rework DEB_BUILD_OPTIONS section
thanks
On Sun, 2002-08-18 at 18:52, James Troup wrote:
I can
guarantee you that the additional time incurred by using '-g' is so
insignificant it's insulting to have to even discuss it.
I
Package: debian-policy
Severity: wishlist
The attached patch should mostly speak for itself. I think it's great
that a lot of packages support DEB_BUILD_OPTIONS=debug now, but if the
program is compiled with optimization, it's very difficult to debug.
Another alternative is to make a whole
On Sat, Aug 17, 2002 at 10:41:11PM -0400, Colin Walters wrote:
Package: debian-policy
Severity: wishlist
The attached patch should mostly speak for itself. I think it's great
that a lot of packages support DEB_BUILD_OPTIONS=debug now, but if the
program is compiled with optimization, it's
On Sat, 17 Aug 2002, Colin Walters wrote:
The attached patch should mostly speak for itself. I think it's great
that a lot of packages support DEB_BUILD_OPTIONS=debug now, but if the
program is compiled with optimization, it's very difficult to debug.
--- debian-policy-3.5.6.1/policy.sgml
34 matches
Mail list logo