-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA224
Slavek, all,
When building TDE on Debian Jessie I have run into several instances where tdebase FTBFS due to the recent changes in CMake policy CMP0046. Setting the OLD behavior repairs the problem.
Are there any drawbacks to setting CMP0046 OLD globally? If not can one of the CMake gurus here please push a proper patch to the cmake module ASAP?
Thanks!
Tim
On 10/06/2014 01:10 PM, Timothy Pearson wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA224
Slavek, all,
When building TDE on Debian Jessie I have run into several instances where tdebase FTBFS due to the recent changes in CMake policy CMP0046. Setting the OLD behavior repairs the problem.
Are there any drawbacks to setting CMP0046 OLD globally? If not can one of the CMake gurus here please push a proper patch to the cmake module ASAP?
Thanks!
Tim -----BEGIN PGP SIGNATURE-----
Tim, I always build on Jessie and haven't run in any FTBFS due to CMP0046 (at least until last Saturday). Nevertheless looking at some log files I can see warnings about that. We can push a patch similar to the one pushed in commit 0d5ca8be for CMP0026. But given the reasons of such warning:
"Error on non-existent dependency in add_dependencies. CMake 2.8.12 and lower silently ignored non-existent dependencies listed in the add_dependencies() command. The OLD behavior for this policy is to silently ignore non-existent dependencies. The NEW behavior for this policy is to report an error if non-existent dependencies are listed in the add_dependencies() command."
IMO it would be better to remove or fix the non-existent dependency.
Perhaps we can push a patch temporarily for v14.0.0 and file a bug report against fixing all non-existent dependencies for after v14.0.0.
What do you think?
Cheers Michele
On Monday 06 of October 2014 06:47:36 Michele Calgaro wrote:
On 10/06/2014 01:10 PM, Timothy Pearson wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA224
Slavek, all,
When building TDE on Debian Jessie I have run into several instances where tdebase FTBFS due to the recent changes in CMake policy CMP0046. Setting the OLD behavior repairs the problem.
Are there any drawbacks to setting CMP0046 OLD globally? If not can one of the CMake gurus here please push a proper patch to the cmake module ASAP?
Thanks!
Tim -----BEGIN PGP SIGNATURE-----
Tim, I always build on Jessie and haven't run in any FTBFS due to CMP0046 (at least until last Saturday). Nevertheless looking at some log files I can see warnings about that. We can push a patch similar to the one pushed in commit 0d5ca8be for CMP0026. But given the reasons of such warning:
"Error on non-existent dependency in add_dependencies. CMake 2.8.12 and lower silently ignored non-existent dependencies listed in the add_dependencies() command. The OLD behavior for this policy is to silently ignore non-existent dependencies. The NEW behavior for this policy is to report an error if non-existent dependencies are listed in the add_dependencies() command."
IMO it would be better to remove or fix the non-existent dependency.
Perhaps we can push a patch temporarily for v14.0.0 and file a bug report against fixing all non-existent dependencies for after v14.0.0.
What do you think?
Cheers Michele
I tried it just now - yes I see many reports CMP0046. However, not caused FTBFS. I'll look at it more closely.
On Monday 06 of October 2014 16:33:30 Slávek Banko wrote:
On Monday 06 of October 2014 06:47:36 Michele Calgaro wrote:
On 10/06/2014 01:10 PM, Timothy Pearson wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA224
Slavek, all,
When building TDE on Debian Jessie I have run into several instances where tdebase FTBFS due to the recent changes in CMake policy CMP0046. Setting the OLD behavior repairs the problem.
Are there any drawbacks to setting CMP0046 OLD globally? If not can one of the CMake gurus here please push a proper patch to the cmake module ASAP?
Thanks!
Tim -----BEGIN PGP SIGNATURE-----
Tim, I always build on Jessie and haven't run in any FTBFS due to CMP0046 (at least until last Saturday). Nevertheless looking at some log files I can see warnings about that. We can push a patch similar to the one pushed in commit 0d5ca8be for CMP0026. But given the reasons of such warning:
"Error on non-existent dependency in add_dependencies. CMake 2.8.12 and lower silently ignored non-existent dependencies listed in the add_dependencies() command. The OLD behavior for this policy is to silently ignore non-existent dependencies. The NEW behavior for this policy is to report an error if non-existent dependencies are listed in the add_dependencies() command."
IMO it would be better to remove or fix the non-existent dependency.
Perhaps we can push a patch temporarily for v14.0.0 and file a bug report against fixing all non-existent dependencies for after v14.0.0.
What do you think?
Cheers Michele
I tried it just now - yes I see many reports CMP0046. However, not caused FTBFS. I'll look at it more closely.
Well, I looked and really this is a bug. For tdebase fixed in commit aaf2ebb5. Setting the CMP0046 to OLD therefore I consider inappropriate. As we discover CMD0046 messages, then should be corrected.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA224
On Monday 06 of October 2014 16:33:30 Slávek Banko wrote:
On Monday 06 of October 2014 06:47:36 Michele Calgaro wrote:
On 10/06/2014 01:10 PM, Timothy Pearson wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA224
Slavek, all,
When building TDE on Debian Jessie I have run into several instances where tdebase FTBFS due to the recent changes in CMake policy
CMP0046.
Setting the OLD behavior repairs the problem.
Are there any drawbacks to setting CMP0046 OLD globally? If not can one of the CMake gurus here please push a proper patch to the cmake module ASAP?
Thanks!
Tim -----BEGIN PGP SIGNATURE-----
Tim, I always build on Jessie and haven't run in any FTBFS due to CMP0046
(at
least until last Saturday). Nevertheless looking at some log files I
can
see warnings about that. We can push a patch similar to the one pushed in commit 0d5ca8be for CMP0026. But given the reasons of such warning:
"Error on non-existent dependency in add_dependencies. CMake 2.8.12 and lower silently ignored non-existent dependencies
listed
in the add_dependencies() command. The OLD behavior for this policy is to silently ignore non-existent dependencies. The NEW behavior for this policy is to report an error
if
non-existent dependencies are listed in the add_dependencies()
command."
IMO it would be better to remove or fix the non-existent dependency.
Perhaps we can push a patch temporarily for v14.0.0 and file a bug report against fixing all non-existent dependencies for after v14.0.0.
What do you think?
Cheers Michele
I tried it just now - yes I see many reports CMP0046. However, not caused FTBFS. I'll look at it more closely.
Well, I looked and really this is a bug. For tdebase fixed in commit aaf2ebb5. Setting the CMP0046 to OLD therefore I consider inappropriate. As we discover CMD0046 messages, then should be corrected.
-- Slávek
Thanks for looking into this. I wasn't sure if the FTBFS was just a glitch on my system.
Tim
On Monday 06 of October 2014 06:10:18 Timothy Pearson wrote:
Slavek, all,
When building TDE on Debian Jessie I have run into several instances where tdebase FTBFS due to the recent changes in CMake policy CMP0046. Setting the OLD behavior repairs the problem.
Are there any drawbacks to setting CMP0046 OLD globally? If not can one of the CMake gurus here please push a proper patch to the cmake module ASAP?
Thanks!
Tim
I watched the all builds if it will occur yet another case CMP0046. I noticed only one - in Konversation. The patch is a single line. Push or wait?
index 5e48cd1..49c5ef1 100644 --- a/konversation/src/linkaddressbook/CMakeLists.txt +++ b/konversation/src/linkaddressbook/CMakeLists.txt @@ -33,6 +33,6 @@ tde_add_library( linkaddressbookui STATIC_PIC AUTOMOC addressbook.cpp nicklisttooltip.cpp nicksonlinetooltip.cpp kimiface.skel addresseeitem.cpp - DEPENDENCIES konversationconfig + DEPENDENCIES konversationconfig-static )
On Thursday 06 of November 2014 00:37:19 Slávek Banko wrote:
On Monday 06 of October 2014 06:10:18 Timothy Pearson wrote:
Slavek, all,
When building TDE on Debian Jessie I have run into several instances where tdebase FTBFS due to the recent changes in CMake policy CMP0046. Setting the OLD behavior repairs the problem.
Are there any drawbacks to setting CMP0046 OLD globally? If not can one of the CMake gurus here please push a proper patch to the cmake module ASAP?
Thanks!
Tim
I watched the all builds if it will occur yet another case CMP0046. I noticed only one - in Konversation. The patch is a single line. Push or wait?
index 5e48cd1..49c5ef1 100644 --- a/konversation/src/linkaddressbook/CMakeLists.txt +++ b/konversation/src/linkaddressbook/CMakeLists.txt @@ -33,6 +33,6 @@ tde_add_library( linkaddressbookui STATIC_PIC AUTOMOC addressbook.cpp nicklisttooltip.cpp nicksonlinetooltip.cpp kimiface.skel addresseeitem.cpp
- DEPENDENCIES konversationconfig
- DEPENDENCIES konversationconfig-static
)
I almost forgot. The patch above solves CMP0046. However, with Konversation occurs even CMP0037. Specifically, due to the localization sr@Latn. This name is used as the name of the target. But, contains invalid characters.
The fix for CMP0037 will not be easy, because the localization files / folders are processed automatically by macro in common cmake module. Therefore, I propose to wait with fix CPM0037 until R14.0.0 will be out of the doors.
What is your opinion?
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA224
On Thursday 06 of November 2014 00:37:19 Slávek Banko wrote:
On Monday 06 of October 2014 06:10:18 Timothy Pearson wrote:
Slavek, all,
When building TDE on Debian Jessie I have run into several instances where tdebase FTBFS due to the recent changes in CMake policy CMP0046. Setting the OLD behavior repairs the problem.
Are there any drawbacks to setting CMP0046 OLD globally? If not can
one
of the CMake gurus here please push a proper patch to the cmake module ASAP?
Thanks!
Tim
I watched the all builds if it will occur yet another case CMP0046. I noticed only one - in Konversation. The patch is a single line. Push or wait?
index 5e48cd1..49c5ef1 100644 --- a/konversation/src/linkaddressbook/CMakeLists.txt +++ b/konversation/src/linkaddressbook/CMakeLists.txt @@ -33,6 +33,6 @@ tde_add_library( linkaddressbookui STATIC_PIC AUTOMOC addressbook.cpp nicklisttooltip.cpp nicksonlinetooltip.cpp kimiface.skel addresseeitem.cpp
- DEPENDENCIES konversationconfig
- DEPENDENCIES konversationconfig-static
)
I almost forgot. The patch above solves CMP0046. However, with Konversation occurs even CMP0037. Specifically, due to the localization sr@Latn. This name is used as the name of the target. But, contains invalid characters.
The fix for CMP0037 will not be easy, because the localization files / folders are processed automatically by macro in common cmake module. Therefore, I propose to wait with fix CPM0037 until R14.0.0 will be out of the doors.
What is your opinion?
Yes, please wait on the CPM0037 fix. Push the CMP0046 fix now.
Thanks!
Tim
-
Michele Calgaro -
Slávek Banko -
Timothy Pearson