Skip to contentSkip to conversation

Autodetect libdrm path (v2)

PATCH In review Tom St Denis 5 participants 10 messages Feb 5, 2017 – Feb 8, 2017 7 messages not in the archive

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Autodetect libdrm path (v2)

(v2): Use findLibDRM script instead of directly finding path

---
 CMakeLists.txt                 |  3 +++
 cmake_modules/FindLibDRM.cmake | 35 +++++++++++++++++++++++++++++++++++
 src/lib/query_drm.c            |  4 ++--
 3 files changed, 40 insertions(+), 2 deletions(-)
 create mode 100644 cmake_modules/FindLibDRM.cmake
CMakeLists.txt +3 -0
index bef94fdba788..ef78c97ad763 100644
@@ -25,6 +25,9 @@ include_directories(${CURSES_INCLUDE_DIRS})
 find_package(PCIAccess REQUIRED)
 include_directories(${PCIACCESS_INCLUDE_DIR})
 
+find_package(LibDRM REQUIRED)
+include_directories(${LIBDRM_INCLUDE_DIR})
+
 set(REQUIRED_EXTERNAL_LIBS
   ${CURSES_LIBRARIES}
   ${PCIACCESS_LIBRARIES}
cmake_modules/FindLibDRM.cmake new file +35 -0
new file mode 100644
index 000000000000..e840c4d1bfd0
@@ -0,0 +1,35 @@
+# Try to find libdrm
+#
+# Once done, this will define
+#
+# LIBDRM_FOUND
+# LIBDRM_INCLUDE_DIR
+# LIBDRM_LIBRARIES
+
+find_package(PkgConfig)
+
+pkg_check_modules(PC_LIBDRM QUIET libdrm)
+
+find_path(LIBDRM_INCLUDE_DIR NAMES amdgpu_drm.h
+    HINTS
+    ${PC_LIBDRM_INCLUDEDIR}
+    ${PC_LIBDRM_INCLUDE_DIRS}
+    /usr/include
+)
+
+find_library(LIBDRM_LIBRARY NAMES libdrm_amdgpu.so.1
+    HINTS
+    ${PC_LIBDRM_LIBDIR}
+    ${PC_LIBDRM_LIBRARY_DIRS}
+    /usr/lib64
+    /usr/lib
+)
+
+SET(LIBDRM_LIBRARIES optimized ${LIBDRM_LIBRARY})
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(LIBDRM DEFAULT_MSG
+	LIBDRM_LIBRARIES LIBDRM_INCLUDE_DIR
+)
+
+mark_as_advanced(LIBDRM_INCLUDE_DIR LIBDRM_LIBRARIES)
src/lib/query_drm.c +2 -2
index b9d80a8fc0c8..755c65fbc662 100644
@@ -25,8 +25,8 @@
 #include "umr.h"
 #include <asm/ioctl.h>
 #include <sys/ioctl.h>
-#include <drm/drm.h>
-#include <drm/amdgpu_drm.h>
+#include <drm.h>
+#include <amdgpu_drm.h>
 
 #define DRM_IOC(dir, group, nr, size) _IOC(dir, group, nr, size)
 #define DRM_IOC_WRITE           _IOC_WRITE
Signature · 6 lines
2.11.0

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): [email protected]

Reviewed-by
#2 lore raw
Quoted 71 lines · Tom St Denis

(v2): Use findLibDRM script instead of directly finding path

---
 CMakeLists.txt                 |  3 +++
 cmake_modules/FindLibDRM.cmake | 35 +++++++++++++++++++++++++++++++++++
 src/lib/query_drm.c            |  4 ++--
 3 files changed, 40 insertions(+), 2 deletions(-)
 create mode 100644 cmake_modules/FindLibDRM.cmake
CMakeLists.txt +3 -0
index bef94fdba788..ef78c97ad763 100644
@@ -25,6 +25,9 @@ include_directories(${CURSES_INCLUDE_DIRS})
 find_package(PCIAccess REQUIRED)
 include_directories(${PCIACCESS_INCLUDE_DIR})
 
+find_package(LibDRM REQUIRED)
+include_directories(${LIBDRM_INCLUDE_DIR})
+
 set(REQUIRED_EXTERNAL_LIBS
   ${CURSES_LIBRARIES}
   ${PCIACCESS_LIBRARIES}
cmake_modules/FindLibDRM.cmake new file +35 -0
new file mode 100644
index 000000000000..e840c4d1bfd0
@@ -0,0 +1,35 @@
+# Try to find libdrm
+#
+# Once done, this will define
+#
+# LIBDRM_FOUND
+# LIBDRM_INCLUDE_DIR
+# LIBDRM_LIBRARIES
+
+find_package(PkgConfig)
+
+pkg_check_modules(PC_LIBDRM QUIET libdrm)
+
+find_path(LIBDRM_INCLUDE_DIR NAMES amdgpu_drm.h
+    HINTS
+    ${PC_LIBDRM_INCLUDEDIR}
+    ${PC_LIBDRM_INCLUDE_DIRS}
+    /usr/include
+)
+
+find_library(LIBDRM_LIBRARY NAMES libdrm_amdgpu.so.1
+    HINTS
+    ${PC_LIBDRM_LIBDIR}
+    ${PC_LIBDRM_LIBRARY_DIRS}
+    /usr/lib64
+    /usr/lib
+)
+
+SET(LIBDRM_LIBRARIES optimized ${LIBDRM_LIBRARY})
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(LIBDRM DEFAULT_MSG
+	LIBDRM_LIBRARIES LIBDRM_INCLUDE_DIR
+)
+
+mark_as_advanced(LIBDRM_INCLUDE_DIR LIBDRM_LIBRARIES)
src/lib/query_drm.c +0 -0
index b9d80a8fc0c8..755c65fbc662 100644

src/lib/query_drm.c line 32

Quoted 10 lines
src/lib/query_drm.c +2 -2
@@ -25,8 +25,8 @@
 #include "umr.h"
 #include <asm/ioctl.h>
 #include <sys/ioctl.h>
-#include <drm/drm.h>
-#include <drm/amdgpu_drm.h>
+#include <drm.h>
+#include <amdgpu_drm.h>
 
 #define DRM_IOC(dir, group, nr, size) _IOC(dir, group, nr, size)
 #define DRM_IOC_WRITE           _IOC_WRITE

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 62ec2a45-ea86-a643-f7fd-3e5e5054540b-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

Reviewed-by Andres Rodriguez
#4 lore raw

Thanks. Pushed it.

Tom

________________________________
From: Andres Rodriguez <[email protected]>
Sent: Sunday, February 5, 2017 17:28
To: Tom St Denis; [email protected]
Cc: StDenis, Tom
Subject: Re: [PATCH] Autodetect libdrm path (v2)

Quoted 71 lines · Tom St Denis

(v2): Use findLibDRM script instead of directly finding path

---
 CMakeLists.txt                 |  3 +++
 cmake_modules/FindLibDRM.cmake | 35 +++++++++++++++++++++++++++++++++++
 src/lib/query_drm.c            |  4 ++--
 3 files changed, 40 insertions(+), 2 deletions(-)
 create mode 100644 cmake_modules/FindLibDRM.cmake
CMakeLists.txt +3 -0
index bef94fdba788..ef78c97ad763 100644
@@ -25,6 +25,9 @@ include_directories(${CURSES_INCLUDE_DIRS})
 find_package(PCIAccess REQUIRED)
 include_directories(${PCIACCESS_INCLUDE_DIR})
 
+find_package(LibDRM REQUIRED)
+include_directories(${LIBDRM_INCLUDE_DIR})
+
 set(REQUIRED_EXTERNAL_LIBS
   ${CURSES_LIBRARIES}
   ${PCIACCESS_LIBRARIES}
cmake_modules/FindLibDRM.cmake new file +35 -0
new file mode 100644
index 000000000000..e840c4d1bfd0
@@ -0,0 +1,35 @@
+# Try to find libdrm
+#
+# Once done, this will define
+#
+# LIBDRM_FOUND
+# LIBDRM_INCLUDE_DIR
+# LIBDRM_LIBRARIES
+
+find_package(PkgConfig)
+
+pkg_check_modules(PC_LIBDRM QUIET libdrm)
+
+find_path(LIBDRM_INCLUDE_DIR NAMES amdgpu_drm.h
+    HINTS
+    ${PC_LIBDRM_INCLUDEDIR}
+    ${PC_LIBDRM_INCLUDE_DIRS}
+    /usr/include
+)
+
+find_library(LIBDRM_LIBRARY NAMES libdrm_amdgpu.so.1
+    HINTS
+    ${PC_LIBDRM_LIBDIR}
+    ${PC_LIBDRM_LIBRARY_DIRS}
+    /usr/lib64
+    /usr/lib
+)
+
+SET(LIBDRM_LIBRARIES optimized ${LIBDRM_LIBRARY})
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(LIBDRM DEFAULT_MSG
+     LIBDRM_LIBRARIES LIBDRM_INCLUDE_DIR
+)
+
+mark_as_advanced(LIBDRM_INCLUDE_DIR LIBDRM_LIBRARIES)
src/lib/query_drm.c +0 -0
index b9d80a8fc0c8..755c65fbc662 100644

src/lib/query_drm.c line 32

Quoted 10 lines
src/lib/query_drm.c +2 -2
@@ -25,8 +25,8 @@
 #include "umr.h"
 #include <asm/ioctl.h>
 #include <sys/ioctl.h>
-#include <drm/drm.h>
-#include <drm/amdgpu_drm.h>
+#include <drm.h>
+#include <amdgpu_drm.h>
 
 #define DRM_IOC(dir, group, nr, size) _IOC(dir, group, nr, size)
 #define DRM_IOC_WRITE           _IOC_WRITE

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

#5 lore raw
In reply to a message not in the archive

Hi Tom,

Quoted 2 lines · Tom St Denis

(v2): Use findLibDRM script instead of directly finding path

Since the project already depends on libdrm you might want to use the drmDevice2 API and drop the iteration over all PCI devices (libpciaccess dependency).
If the former is lacking something feel free to suggest/send patches ;-)

The libdrm requirement was missing last time I've skimmed through the README.

Quoted 7 lines

---

cmake_modules/FindLibDRM.cmake new file +1 -0
@@ -0,0 +1,35 @@
+# Try to find libdrm

cmake_modules/FindLibDRM.cmake line 11

Quoted 10 lines
cmake_modules/FindLibDRM.cmake new file +10 -0
+#
+# Once done, this will define
+#
+# LIBDRM_FOUND
+# LIBDRM_INCLUDE_DIR
+# LIBDRM_LIBRARIES
+
+find_package(PkgConfig)
+
+pkg_check_modules(PC_LIBDRM QUIET libdrm)

You really want a version check.

Quoted 6 lines
cmake_modules/FindLibDRM.cmake new file +6 -0
+
+find_path(LIBDRM_INCLUDE_DIR NAMES amdgpu_drm.h
+    HINTS
+    ${PC_LIBDRM_INCLUDEDIR}
+    ${PC_LIBDRM_INCLUDE_DIRS}
+    /usr/include

cmake_modules/FindLibDRM.cmake line 27

Quoted 10 lines
cmake_modules/FindLibDRM.cmake new file +10 -0
+)
+
+find_library(LIBDRM_LIBRARY NAMES libdrm_amdgpu.so.1
+    HINTS
+    ${PC_LIBDRM_LIBDIR}
+    ${PC_LIBDRM_LIBRARY_DIRS}
+    /usr/lib64
+    /usr/lib
+)
+

Here is one of the things which makes me rage against cmake - no picking /usr/{include,foo} is _not_ what you want.

During cross-compilation as one is missing the .pc file, the system headers/libraries will be picked. This is horribly wrong, yet rather common in cmake world.
What you really want is a lovely error message to warn the user - s/OPTIONAL/REQUIRED/ will give you just that.

At which point, checking for the header/library in the paths provided by the .pc file is redundant and thus nearly everything else in the file ;-)

TL;Dr: All you need is pkg_check_modules(PC_LIBDRM REQUIRED libdrm>=XX).

Thanks
Emil
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): CACvgo5230Use5sk8dm5jBbLqiJeSkgB6cDikz-N4omx-4Jgy9A-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org

#7 lore raw

I have to NAK that idea since we use umr on NPI work which doesn't necessarily have libdrm support.

Also umr can read/write registers via pci access without amdgpu loaded (handy if amdgpu fails to init properly).

Though you are right that libdrm is technically a requirement and I should add that to the README.

Tom
________________________________
From: Emil Velikov <[email protected]>
Sent: Monday, February 6, 2017 15:18
To: Tom St Denis
Cc: amd-gfx mailing list; StDenis, Tom
Subject: Re: [PATCH] Autodetect libdrm path (v2)

Hi Tom,

Quoted 2 lines · Tom St Denis

(v2): Use findLibDRM script instead of directly finding path

Since the project already depends on libdrm you might want to use the drmDevice2 API and drop the iteration over all PCI devices (libpciaccess dependency).
If the former is lacking something feel free to suggest/send patches ;-)

The libdrm requirement was missing last time I've skimmed through the README.

Quoted 7 lines

---

cmake_modules/FindLibDRM.cmake new file +1 -0
@@ -0,0 +1,35 @@
+# Try to find libdrm

cmake_modules/FindLibDRM.cmake line 11

Quoted 10 lines
cmake_modules/FindLibDRM.cmake new file +10 -0
+#
+# Once done, this will define
+#
+# LIBDRM_FOUND
+# LIBDRM_INCLUDE_DIR
+# LIBDRM_LIBRARIES
+
+find_package(PkgConfig)
+
+pkg_check_modules(PC_LIBDRM QUIET libdrm)

You really want a version check.

Quoted 6 lines
cmake_modules/FindLibDRM.cmake new file +6 -0
+
+find_path(LIBDRM_INCLUDE_DIR NAMES amdgpu_drm.h
+    HINTS
+    ${PC_LIBDRM_INCLUDEDIR}
+    ${PC_LIBDRM_INCLUDE_DIRS}
+    /usr/include

cmake_modules/FindLibDRM.cmake line 27

Quoted 10 lines
cmake_modules/FindLibDRM.cmake new file +10 -0
+)
+
+find_library(LIBDRM_LIBRARY NAMES libdrm_amdgpu.so.1
+    HINTS
+    ${PC_LIBDRM_LIBDIR}
+    ${PC_LIBDRM_LIBRARY_DIRS}
+    /usr/lib64
+    /usr/lib
+)
+

Here is one of the things which makes me rage against cmake - no picking /usr/{include,foo} is _not_ what you want.

During cross-compilation as one is missing the .pc file, the system headers/libraries will be picked. This is horribly wrong, yet rather common in cmake world.
What you really want is a lovely error message to warn the user - s/OPTIONAL/REQUIRED/ will give you just that.

At which point, checking for the header/library in the paths provided by the .pc file is redundant and thus nearly everything else in the file ;-)

TL;Dr: All you need is pkg_check_modules(PC_LIBDRM REQUIRED libdrm>=XX).

Thanks
Emil

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): CY4PR12MB17689AEC15DEF9662A16DEA3F7400-rpdhrqHFk06yjjPBNVDk/[email protected]

#9 lore raw

Apparently I missed the bottom of your reply (all of the clients I have outlook/gmail do top post only ...) as for the cmake changes. I'm for it if you want to submit a patch. I can't imagine a lot of cross compiling though since users will typically be using it on the platform they built it for.

Ironically, I had the pkg_check originally but was told that's a faux-pas.

Unless this is breaking for actual users though it's not really a priority to bikeshed the build system of a 30 file project that is meant to work only on developer machines who are likely building for themselves [😊]

Tom

________________________________
From: amd-gfx <[email protected]> on behalf of StDenis, Tom <[email protected]>
Sent: Monday, February 6, 2017 16:33
To: Emil Velikov; Tom St Denis
Cc: amd-gfx mailing list
Subject: Re: [PATCH] Autodetect libdrm path (v2)

I have to NAK that idea since we use umr on NPI work which doesn't necessarily have libdrm support.

Also umr can read/write registers via pci access without amdgpu loaded (handy if amdgpu fails to init properly).

Though you are right that libdrm is technically a requirement and I should add that to the README.

Tom
________________________________
From: Emil Velikov <[email protected]>
Sent: Monday, February 6, 2017 15:18
To: Tom St Denis
Cc: amd-gfx mailing list; StDenis, Tom
Subject: Re: [PATCH] Autodetect libdrm path (v2)

Hi Tom,

Quoted 2 lines · Tom St Denis

(v2): Use findLibDRM script instead of directly finding path

Since the project already depends on libdrm you might want to use the drmDevice2 API and drop the iteration over all PCI devices (libpciaccess dependency).
If the former is lacking something feel free to suggest/send patches ;-)

The libdrm requirement was missing last time I've skimmed through the README.

Quoted 7 lines

---

cmake_modules/FindLibDRM.cmake new file +1 -0
@@ -0,0 +1,35 @@
+# Try to find libdrm

cmake_modules/FindLibDRM.cmake line 11

Quoted 10 lines
cmake_modules/FindLibDRM.cmake new file +10 -0
+#
+# Once done, this will define
+#
+# LIBDRM_FOUND
+# LIBDRM_INCLUDE_DIR
+# LIBDRM_LIBRARIES
+
+find_package(PkgConfig)
+
+pkg_check_modules(PC_LIBDRM QUIET libdrm)

You really want a version check.

Quoted 6 lines
cmake_modules/FindLibDRM.cmake new file +6 -0
+
+find_path(LIBDRM_INCLUDE_DIR NAMES amdgpu_drm.h
+    HINTS
+    ${PC_LIBDRM_INCLUDEDIR}
+    ${PC_LIBDRM_INCLUDE_DIRS}
+    /usr/include

cmake_modules/FindLibDRM.cmake line 27

Quoted 10 lines
cmake_modules/FindLibDRM.cmake new file +10 -0
+)
+
+find_library(LIBDRM_LIBRARY NAMES libdrm_amdgpu.so.1
+    HINTS
+    ${PC_LIBDRM_LIBDIR}
+    ${PC_LIBDRM_LIBRARY_DIRS}
+    /usr/lib64
+    /usr/lib
+)
+

Here is one of the things which makes me rage against cmake - no picking /usr/{include,foo} is _not_ what you want.

During cross-compilation as one is missing the .pc file, the system headers/libraries will be picked. This is horribly wrong, yet rather common in cmake world.
What you really want is a lovely error message to warn the user - s/OPTIONAL/REQUIRED/ will give you just that.

At which point, checking for the header/library in the paths provided by the .pc file is redundant and thus nearly everything else in the file ;-)

TL;Dr: All you need is pkg_check_modules(PC_LIBDRM REQUIRED libdrm>=XX).

Thanks
Emil

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): CY4PR12MB1768AED537A0253EB70B1E4BF7400-rpdhrqHFk06yjjPBNVDk/[email protected]

#11 lore raw
Quoted 2 lines · Tom

Apparently I missed the bottom of your reply (all of the clients I have outlook/gmail do top post only ...)

Both Outlook and Gmail can do inline replies and plain text. There might be some magic required for the former though :-\
I would kindly suggest using inline/text when possible.

Quoted 5 lines

as for the cmake changes. I'm for it if you want to submit a patch. I can't imagine a lot of cross compiling though since users will typically be using it on the platform they built it for.

Ironically, I had the pkg_check originally but was told that's a faux-pas.

It's a common misconception, influenced by the sheer volume of copy/paste :-)

Quoted 3 lines

Unless this is breaking for actual users though it's not really a priority to bikeshed the build system of a 30 file project that is meant to work only on developer machines who are likely building for themselves

Up-to you really. FWIW using cmake/autoconf/etc. is a huge overkill. The original makefile was missing a few things* worth 10-20 lines of code while being noticeably smaller, quicker and easier to read ;-)

* Dependency tracking (pkg-config + foo.pc), incremental builds - ($CC ... -MM -MP ...), tarballs.

Quoted 8 lines

Tom

________________________________
From: amd-gfx <[email protected]> on behalf of StDenis, Tom <[email protected]>
Sent: Monday, February 6, 2017 16:33
To: Emil Velikov; Tom St Denis
Cc: amd-gfx mailing list

Quoted 8 lines

Subject: Re: [PATCH] Autodetect libdrm path (v2)

I have to NAK that idea since we use umr on NPI work which doesn't necessarily have libdrm support.

Also umr can read/write registers via pci access without amdgpu loaded (handy if amdgpu fails to init properly).

Must have missed this part. Thanks for the correction.

Quoted 3 lines

Though you are right that libdrm is technically a requirement and I should add that to the README.

Ack.

Thanks
Emil
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): CACvgo537ciLvg=Q6enktu6ndMSEuSCM+RpmniJdgYNgZDQjUeQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org

#13 lore raw
Quoted 6 lines · Emil Velikov
Quoted 2 lines · Tom

Apparently I missed the bottom of your reply (all of the clients I have outlook/gmail do top post only ...)

Both Outlook and Gmail can do inline replies and plain text. There might be some magic required for the former though :-\
I would kindly suggest using inline/text when possible.

You'll be pleased to note I discovered (at some point) that AMD has re-enabled imap support and I can now use a more sensible client...

:-)

Quoted 10 lines
Quoted 2 lines

Ironically, I had the pkg_check originally but was told that's a faux-pas.

It's a common misconception, influenced by the sheer volume of copy/paste :-)

Quoted 3 lines

Unless this is breaking for actual users though it's not really a priority to bikeshed the build system of a 30 file project that is meant to work only on developer machines who are likely building for themselves

Up-to you really. FWIW using cmake/autoconf/etc. is a huge overkill. The original makefile was missing a few things* worth 10-20 lines of code while being noticeably smaller, quicker and easier to read ;-)

To be honest unless the cmake system gets in the way I'm happy if it makes others happy. It's not easy to say the project is "open" if we NAK any and all harmless commits that the users advocate for.

The official stance from AMD (specifically the team I work for) is if cmake gets in the way of AMD work we will nuke the cmake system and go back to gmake (with pkg-config shell calls to get paths).

Personally I'm not a big fan of overly complicated config/build systems but since I don't do package building/etc for a living I'll defer to the opinions of others within reason.

Quoted 3 lines
Quoted 2 lines

Also umr can read/write registers via pci access without amdgpu loaded (handy if amdgpu fails to init properly).

Must have missed this part. Thanks for the correction.

Yup, you can force device + direct access, e.g.

umr -O use_pci -f fiji -r *.uvd5.mmFOO

Which is handy like I said if the module init fails. Developers also use it in NPI bringup to poke things.

Cheers,
Tom
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 13f9479c-1a5c-8887-858e-dadf819a7140-5C7GfCeVMHo@public.gmane.org

#15 lore raw
Quoted 13 lines · Tom St Denis
Quoted 10 lines · Emil Velikov
Quoted 4 lines · Tom

Apparently I missed the bottom of your reply (all of the clients I have outlook/gmail do top post only ...)

Both Outlook and Gmail can do inline replies and plain text. There might be some magic required for the former though :-\
I would kindly suggest using inline/text when possible.

Quoted 5 lines

You'll be pleased to note I discovered (at some point) that AMD has re-enabled imap support and I can now use a more sensible client...

:-)

Amazing news, thank you !

Quoted 20 lines
Quoted 7 lines
Quoted 3 lines

Ironically, I had the pkg_check originally but was told that's a faux-pas.

It's a common misconception, influenced by the sheer volume of copy/paste :-)

Quoted 1 line
Quoted 7 lines
Quoted 4 lines

Unless this is breaking for actual users though it's not really a priority to bikeshed the build system of a 30 file project that is meant to work only on developer machines who are likely building for themselves

Up-to you really. FWIW using cmake/autoconf/etc. is a huge overkill. The original makefile was missing a few things* worth 10-20 lines of code while being noticeably smaller, quicker and easier to read ;-)

To be honest unless the cmake system gets in the way I'm happy if it makes others happy. It's not easy to say the project is "open" if we NAK any and all harmless commits that the users advocate for.

Quoted 8 lines

The official stance from AMD (specifically the team I work for) is if cmake gets in the way of AMD work we will nuke the cmake system and go back to gmake (with pkg-config shell calls to get paths).

Personally I'm not a big fan of overly complicated config/build systems but since I don't do package building/etc for a living I'll defer to the opinions of others within reason.

Bth, you're not missing anything much by not tinkering with autoconf/cmake/etc. It's just a bikeshed fest with overly biased people on each side ;-)

Thanks
Emil
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

#16 lore raw
In reply to a message not in the archive
Quoted 11 lines · Tom St Denis
Quoted 7 lines · Emil Velikov
Quoted 3 lines · Tom

Apparently I missed the bottom of your reply (all of the clients I have outlook/gmail do top post only ...)

Both Outlook and Gmail can do inline replies and plain text. There might be some magic required for the former though :-\
I would kindly suggest using inline/text when possible.

You'll be pleased to note I discovered (at some point) that AMD has re-enabled imap support

FWIW, I've always used IMAP for AMD e-mail, it was never disabled per se.

Quoted 3 lines

and I can now use a more sensible client...

:-)

Anyway, that's great.

Signature · 6 lines
Earthling Michel Dänzer               |               http://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx