The MythTV lists have had several reports of problems apparently arising from new (and improved) vector-validation in gcc8. Is there a temporary workaround?
It may be a coincidence, but the MythTV bug-tracker is currently reporting a temporary and apparently Fedora-related error:
Something has triggered an error on your website. This is the default error page for nginx that is distributed with Fedora. It is located /usr/share/nginx/html/50x.html
On Mon, May 07, 2018 at 09:51:52AM +0100, John Pilkington wrote:
The MythTV lists have had several reports of problems apparently arising from new (and improved) vector-validation in gcc8. Is there a temporary workaround?
If you don't provide details (links to the reports, etc.), then it is impossible to tell anything.
Jakub
On 07/05/18 10:13, Jakub Jelinek wrote:
On Mon, May 07, 2018 at 09:51:52AM +0100, John Pilkington wrote:
The MythTV lists have had several reports of problems apparently arising from new (and improved) vector-validation in gcc8. Is there a temporary workaround?
If you don't provide details (links to the reports, etc.), then it is impossible to tell anything.
Jakub
As I said, the bug-tracker is currently out of action. Here's the original report:
http://lists.mythtv.org/pipermail/mythtv-commits/2018-May/117389.html
On Mon, May 07, 2018 at 10:40:39AM +0100, John Pilkington wrote:
On 07/05/18 10:13, Jakub Jelinek wrote:
On Mon, May 07, 2018 at 09:51:52AM +0100, John Pilkington wrote:
The MythTV lists have had several reports of problems apparently arising from new (and improved) vector-validation in gcc8. Is there a temporary workaround?
If you don't provide details (links to the reports, etc.), then it is impossible to tell anything.
Jakub
As I said, the bug-tracker is currently out of action. Here's the original report:
http://lists.mythtv.org/pipermail/mythtv-commits/2018-May/117389.html
If mythtv invoked UB, then you really just want to fix the UB, not look for workarounds.
Jakub
On 07/05/18 12:14, Jakub Jelinek wrote:
On Mon, May 07, 2018 at 10:40:39AM +0100, John Pilkington wrote:
On 07/05/18 10:13, Jakub Jelinek wrote:
On Mon, May 07, 2018 at 09:51:52AM +0100, John Pilkington wrote:
The MythTV lists have had several reports of problems apparently arising from new (and improved) vector-validation in gcc8. Is there a temporary workaround?
If you don't provide details (links to the reports, etc.), then it is impossible to tell anything.
Jakub
As I said, the bug-tracker is currently out of action. Here's the original report:
http://lists.mythtv.org/pipermail/mythtv-commits/2018-May/117389.html
If mythtv invoked UB, then you really just want to fix the UB, not look for workarounds.
Jakub
Agreed. But it seems unlikely that MythTV will be the only programme affected, and people who rely on packaged builds won't immediately have that option. I'm still on F26...and SL7
Il giorno lun 7 mag 2018 alle 13:28, John Pilkington johnpilk222@gmail.com ha scritto:
On 07/05/18 12:14, Jakub Jelinek wrote:
On Mon, May 07, 2018 at 10:40:39AM +0100, John Pilkington wrote:
On 07/05/18 10:13, Jakub Jelinek wrote:
On Mon, May 07, 2018 at 09:51:52AM +0100, John Pilkington wrote:
The MythTV lists have had several reports of problems apparently arising from new (and improved) vector-validation in gcc8. Is there a temporary workaround?
If you don't provide details (links to the reports, etc.), then it is impossible to tell anything.
Jakub
As I said, the bug-tracker is currently out of action. Here's the original report:
http://lists.mythtv.org/pipermail/mythtv-commits/2018-May/117389.html
If mythtv invoked UB, then you really just want to fix the UB, not look for workarounds.
Jakub
Agreed. But it seems unlikely that MythTV will be the only programme affected, and people who rely on packaged builds won't immediately have that option. I'm still on F26...and SL7
In fact, lilypond is another package affected: https://bugzilla.redhat.com/show_bug.cgi?id=1568274 http://lists.gnu.org/archive/html/lilypond-user/2018-05/msg00081.html
Federico Bruni writes:
In fact, lilypond is another package affected: https://bugzilla.redhat.com/show_bug.cgi?id=1568274 http://lists.gnu.org/archive/html/lilypond-user/2018-05/msg00081.html
It's been my experience, over the last decade, or so, that each successive major gcc release has had much less tolerance for undefined behavior. And that the correct approach is to fix the undefined behavior. So, if lilypond is triggering undefined behavior, then that's what needs to be fixed, not gcc.
-
Federico Bruni -
Jakub Jelinek -
John Pilkington -
Sam Varshavchik