Skip to contentSkip to conversation

ncurses: version bump to 5.8

PATCH New Mike Frysinger 4 participants 20 messages Mar 23, 2011 – Apr 22, 2011

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ncurses: version bump to 5.8

---
note: not really tested since external toolchain support is totally broken

 package/ncurses/ncurses.mk |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
package/ncurses/ncurses.mk +1 -1
index 0a49887..362f3ed 100644
@@ -23,7 +23,7 @@
 # USA
 
 # TARGETS
-NCURSES_VERSION = 5.7
+NCURSES_VERSION = 5.8
 NCURSES_SITE = $(BR2_GNU_MIRROR)/ncurses
 NCURSES_SOURCE = ncurses-$(NCURSES_VERSION).tar.gz
 NCURSES_INSTALL_STAGING = YES
Signature · 1 line
1.7.4.1
#1 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mike" == Mike Frysinger <[email protected]> writes:

Mike> Signed-off-by: Mike Frysinger <[email protected]>

Committed, thanks.

Mike> note: not really tested since external toolchain support is Mike> totally broken

Ohh?

Signature · 1 line
Bye, Peter Korsgaard
#2 lore raw

PATCH 1/1 ncurses: version bump to 5.8

In reply to Mike Frysinger
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mike" == Mike Frysinger <[email protected]> writes:

Mike> Signed-off-by: Mike Frysinger <[email protected]>
Mike> ---
Mike> note: not really tested since external toolchain support is totally broken

Have you had a chance to test it now? It seems to be crashing on all the configs I've tried (arm9 / x86, gcc 4.3.x / 4.5.x, uClibc 0.9.31 / 0.9.32-rc3).

The problem is a null pointer access in tgetent. It seems to load the terminfo file ok, but something goes wrong.

Reverting back to 5.7 everything works fine aagain.

 Mike>  package/ncurses/ncurses.mk |    2 +-

Mike> 1 files changed, 1 insertions(+), 1 deletions(-)

Mike> diff --git a/package/ncurses/ncurses.mk b/package/ncurses/ncurses.mk Mike> index 0a49887..362f3ed 100644
Mike> --- a/package/ncurses/ncurses.mk
Mike> +++ b/package/ncurses/ncurses.mk
Mike> @@ -23,7 +23,7 @@
Mike> # USA

Mike> # TARGETS
Mike> -NCURSES_VERSION = 5.7
Mike> +NCURSES_VERSION = 5.8

 Mike>  NCURSES_SITE = $(BR2_GNU_MIRROR)/ncurses
 Mike>  NCURSES_SOURCE = ncurses-$(NCURSES_VERSION).tar.gz
 Mike>  NCURSES_INSTALL_STAGING = YES

Mike> --
Mike> 1.7.4.1

Signature · 1 line
Bye, Peter Korsgaard
#3 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 6 lines · Peter Korsgaard
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mike" == Mike Frysinger writes:

?Mike> Signed-off-by: Mike Frysinger <[email protected]>
?Mike> ---
?Mike> note: not really tested since external toolchain support is totally broken

Have you had a chance to test it now?

ive tested external toolchain support and it works fine for me now after merging Sonic's patch

as for ncurses-5.8 itself, i deployed it in Gentoo about a month ago and no one has filed a bug about it ...
-mike

#4 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mike" == Mike Frysinger <[email protected]> writes:

Hi,

Quoted 1 line
Quoted 1 line

Have you had a chance to test it now?

Mike> ive tested external toolchain support and it works fine for me now Mike> after merging Sonic's patch

Ok.

Mike> as for ncurses-5.8 itself, i deployed it in Gentoo about a month ago Mike> and no one has filed a bug about it ...

Ok, odd. I do see several gentoo bugs related to ncurses 5.8 though (rtorrent, vifm).

Signature · 1 line
Bye, Peter Korsgaard
#5 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 6 lines · Peter Korsgaard
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mike" == Mike Frysinger writes:

?Mike> as for ncurses-5.8 itself, i deployed it in Gentoo about a month ago ?Mike> and no one has filed a bug about it ...

Ok, odd. I do see several gentoo bugs related to ncurses 5.8 though (rtorrent, vifm).

seems like all the logged gentoo bugs are issues in the packages and not ncurses

did you have a particular app you were running from BR that exhibited the issue ?
-mike

#6 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mike" == Mike Frysinger <[email protected]> writes:

Hi,

Quoted 2 lines
Quoted 2 lines

Ok, odd. I do see several gentoo bugs related to ncurses 5.8 though (rtorrent, vifm).

Mike> seems like all the logged gentoo bugs are issues in the packages Mike> and not ncurses

Somewhat true. Looking at the Debian bugs I see ncurses 5.8 fails on newwin(0, 0, 0, 0) which it shouldn't - That's fixed post 5.8 though.

Mike> did you have a particular app you were running from BR that Mike> exhibited the issue ?

Well, everything I tried:

- python with it's readline module
- less
- dialog

Signature · 1 line
Bye, Peter Korsgaard
#7 lore raw

PATCH 1/1 ncurses: version bump to 5.8

In reply to Peter Korsgaard
Quoted 2 lines · Peter Korsgaard

The problem is a null pointer access in tgetent. It seems to load the terminfo file ok, but something goes wrong.

https://bugs.busybox.net/show_bug.cgi?id=2845

we shoun't build curses with --with-termlib=tinfo

I have made a draft of a patch
https://bugs.busybox.net/show_bug.cgi?id=2845#c8

Signature · 8 lines
S?rgio M. B.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3309 bytes
Desc: not available
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20110404/2cf4f18c/attachment.bin>
#8 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Sergio" == Sergio Monteiro Basto <[email protected]> writes:

Quoted 2 lines · Peter Korsgaard
Quoted 2 lines

The problem is a null pointer access in tgetent. It seems to load the terminfo file ok, but something goes wrong.

Sergio> https://bugs.busybox.net/show_bug.cgi?id=2845

Sergio> we shoun't build curses with --with-termlib=tinfo

Does that have anything to do with the crash? Your bug report (which I cannot reproduce) seems to be related to a build error rather than a runtime one.

I've now just tested with a glibc toolchain (E.G qemu_arm_versatile_defconfig and select external codesourcery 2010.09 toolchain) and stuff using ncurses (I tried htop) also crashes here, so it doesn't seem to be something specific to uclibc.

I've also tested ncurses 5.9 with the same results. with 2011.05 getting quite near, perhaps the best approach is to revert to ncurses 5.7?

Signature · 1 line
Bye, Peter Korsgaard
#9 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 6 lines · Peter Korsgaard
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Sergio" == Sergio Monteiro Basto <[email protected]> writes:

Quoted 2 lines · Peter Korsgaard
Quoted 2 lines

The problem is a null pointer access in tgetent. It seems to load the terminfo file ok, but something goes wrong.

Quoted 7 lines

Sergio> https://bugs.busybox.net/show_bug.cgi?id=2845

Sergio> we shoun't build curses with --with-termlib=tinfo

Does that have anything to do with the crash? Your bug report (which I cannot reproduce) seems to be related to a build error rather than a runtime one.

https://bugs.busybox.net/show_bug.cgi?id=2845#c7

After you fix the build error, I wrote that we should built ncurses with configure option --with-termlib=tinfo , which Fedora use by default . And you wrote to make a "patch adding optional tinfo build" , I done one draft of the patch.

Since this bug in runtime, seems to be with terminfo, build with option --with-termlib=tinfo , doesn't solve the problem ?

Quoted 8 lines

I've now just tested with a glibc toolchain (E.G qemu_arm_versatile_defconfig and select external codesourcery 2010.09 toolchain) and stuff using ncurses (I tried htop) also crashes here, so it doesn't seem to be something specific to uclibc.

I've also tested ncurses 5.9 with the same results. with 2011.05 getting quite near, perhaps the best approach is to revert to ncurses 5.7?

Signature · 8 lines
S?rgio M. B.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3309 bytes
Desc: not available
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20110420/301f26a2/attachment.bin>
#10 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Sergio" == Sergio Monteiro Basto <[email protected]> writes:

Hi,

Sergio> After you fix the build error, I wrote that we should built Sergio> ncurses with configure option --with-termlib=tinfo , which Sergio> Fedora use by default . And you wrote to make a "patch adding Sergio> optional tinfo build" , I done one draft of the patch.

Sergio> Since this bug in runtime, seems to be with terminfo, build with option Sergio> --with-termlib=tinfo , doesn't solve the problem ?

I haven't tried, but doesn't that just shuffle part of the ncurses functions into libtinfo? I don't right away see why that would fix anything, but I'll start a build now.

Besides this hypothetical fix, is there any other reason why --with-termlib=tinfo would be preferable?

Signature · 1 line
Bye, Peter Korsgaard
#11 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 9 lines · Peter Korsgaard
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Sergio" == Sergio Monteiro Basto <[email protected]> writes:

Hi,

Sergio> After you fix the build error, I wrote that we should built Sergio> ncurses with configure option --with-termlib=tinfo , which Sergio> Fedora use by default . And you wrote to make a "patch adding Sergio> optional tinfo build" , I done one draft of the patch.

Quoted 6 lines

Sergio> Since this bug in runtime, seems to be with terminfo, build with option Sergio> --with-termlib=tinfo , doesn't solve the problem ?

I haven't tried, but doesn't that just shuffle part of the ncurses functions into libtinfo? I don't right away see why that would fix anything, but I'll start a build now.

so, it fix anything ?

Quoted 2 lines

Besides this hypothetical fix, is there any other reason why --with-termlib=tinfo would be preferable?

Well, What I know is.
Is a default of Fedora and seems to me that ncurses becomes better. In fact builds libtinfo.so which will be use (when exist) in some others packages, one of them is dialog.
I can't say nothing more for sure , but packages use libtinfo.so if they exist by default, so probably will work better if lib exist.

Best regards,

Signature · 8 lines
S?rgio M. B.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3309 bytes
Desc: not available
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20110421/16a29976/attachment.bin>
#12 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Sergio" == Sergio Monteiro Basto <[email protected]> writes:

Hi,

Quoted 3 lines
Quoted 3 lines

I haven't tried, but doesn't that just shuffle part of the ncurses functions into libtinfo? I don't right away see why that would fix anything, but I'll start a build now.

Sergio> so, it fix anything ?

No, as expected the behaviour is exactly the same.

Signature · 1 line
Bye, Peter Korsgaard
#13 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 8 lines · Peter Korsgaard
Quoted 6 lines
Quoted 3 lines · Sergio Monteiro Basto

I haven't tried, but doesn't that just shuffle part of the ncurses functions into libtinfo? I don't right away see why that would fix anything, but I'll start a build now.

so, it fix anything ?

No, as expected the behaviour is exactly the same.

ncurses 5.9 is out and seems to be mostly a bugfix release over 5.8 ... probably worth testing
-mike

#14 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 6 lines · Mike Frysinger
Quoted 5 lines · Peter Korsgaard
Quoted 5 lines
Quoted 3 lines · Sergio Monteiro Basto

I haven't tried, but doesn't that just shuffle part of the ncurses functions into libtinfo? I don't right away see why that would fix anything, but I'll start a build now.

Quoted 6 lines
Quoted 3 lines
Quoted 1 line

so, it fix anything ?

No, as expected the behaviour is exactly the same.

ncurses 5.9 is out and seems to be mostly a bugfix release over 5.8 ... probably worth testing

In this thread Peter was wrote:

"I've also tested ncurses 5.9 with the same results. with 2011.05 getting
quite near, perhaps the best approach is to revert to ncurses 5.7? "

Best regards,

Signature · 8 lines
S?rgio M. B.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3309 bytes
Desc: not available
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20110422/d91bb433/attachment.bin>
#15 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 2 lines · Sergio Monteiro Basto

"I've also tested ncurses 5.9 with the same results. with 2011.05 getting quite near, perhaps the best approach is to revert to ncurses 5.7?"

probably reverting to 5.7 for the release makes sense -mike

#16 lore raw

PATCH 1/1 ncurses: version bump to 5.8

In reply to Mike Frysinger
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mike" == Mike Frysinger <[email protected]> writes:

Hi,

Mike> ncurses 5.9 is out and seems to be mostly a bugfix release over 5.8 Mike> ... probably worth testing

As I mentioned in my previous mail, I already tried 5.9 and got same crash as on 5.8 (arm926, both on glibc and uclibc).

Signature · 1 line
Bye, Peter Korsgaard
#17 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoting Peter Korsgaard <[email protected]>:

Quoted 9 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mike" == Mike Frysinger <[email protected]> writes:

Hi,

Mike> ncurses 5.9 is out and seems to be mostly a bugfix release over 5.8 Mike> ... probably worth testing

As I mentioned in my previous mail, I already tried 5.9 and got same crash as on 5.8 (arm926, both on glibc and uclibc).

The problem has something to do with readline.
I tested it against bash and as soon as readline support is off, the seg fault disappears.

Signature · 7 lines
Regards
   Heiko Zuerker
   http://www.devil-linux.org


----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
#18 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoted 4 lines
Quoted 4 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Heiko" == Heiko Zuerker <[email protected]> writes:

As I mentioned in my previous mail, I already tried 5.9 and got same crash as on 5.8 (arm926, both on glibc and uclibc).

Heiko> The problem has something to do with readline.
Heiko> I tested it against bash and as soon as readline support is off, the Heiko> seg fault disappears.

But if you build without readline support, then bash isn't linked against ncurses at all, is it?

Signature · 1 line
Bye, Peter Korsgaard
#19 lore raw

PATCH 1/1 ncurses: version bump to 5.8

Quoting Peter Korsgaard <[email protected]>:

Quoted 11 lines
Quoted 4 lines
Quoted 4 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Heiko" == Heiko Zuerker <[email protected]> writes:

As I mentioned in my previous mail, I already tried 5.9 and got same crash as on 5.8 (arm926, both on glibc and uclibc).

Heiko> The problem has something to do with readline.
Heiko> I tested it against bash and as soon as readline support is off, the Heiko> seg fault disappears.

But if you build without readline support, then bash isn't linked against ncurses at all, is it?

I did a quick test compile and you're right. I'll take back what I said and blame ncurses again. ;-)

Signature · 7 lines
Regards
   Heiko Zuerker
   http://www.devil-linux.org


----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.