New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Remove ./configure script #73

Closed

ChrisCummins opened this issue Jan 13, 2020 · 1 comment

Assignees

Owner

ChrisCummins commented Jan 13, 2020

If tensorflow-gpu dep is broken (which it appears to be, see #65), then we may as well drop the configure script.

The text was updated successfully, but these errors were encountered:

ChrisCummins self-assigned this

ChrisCummins added a commit that referenced this issue


          configure: Split git submodule logic into separate script.

e9a90b8

This pulls out the logic for updating and rewriting git submodules
into a separate script at ./tools/update_git_submodules.py.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Pull out bootstrapping into separate script.

df03204

This moves the ./bootstrap.sh script creation into a seaparate script
./tools/bootstrap.py.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Remove accrued dead code.

abcc295

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Remove --with_cuda option.

7e8f19f

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          Remove ./configure script.

ef18664

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Split git submodule logic into separate script.

2e6c69d

This pulls out the logic for updating and rewriting git submodules
into a separate script at ./tools/update_git_submodules.py.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Pull out bootstrapping into separate script.

841c9b5

This moves the ./bootstrap.sh script creation into a seaparate script
./tools/bootstrap.py.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Remove accrued dead code.

6b18f5e

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Remove --with_cuda option.

19bc330

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          Remove ./configure script.

7aed4a7

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Split git submodule logic into separate script.

db31046

This pulls out the logic for updating and rewriting git submodules
into a separate script at ./tools/update_git_submodules.py.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Pull out bootstrapping into separate script.

adb28de

This moves the ./bootstrap.sh script creation into a seaparate script
./tools/bootstrap.py.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Remove accrued dead code.

a198a03

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Remove --with_cuda option.

b4de09c

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          Remove ./configure script.

bc6416b

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Split git submodule logic into separate script.

6e643ba

This pulls out the logic for updating and rewriting git submodules
into a separate script at ./tools/update_git_submodules.py.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Pull out bootstrapping into separate script.

b1587d1

This moves the ./bootstrap.sh script creation into a seaparate script
./tools/bootstrap.py.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Remove accrued dead code.

f2d9caa

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          configure: Remove --with_cuda option.

65210d5

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          Remove ./configure script.

0083dd8

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com//issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit that referenced this issue


          Remove ./configure script

8670ed1

github.com//issues/73

Owner Author

ChrisCummins commented Jan 24, 2020

Done.

ChrisCummins closed this as completed

ChrisCummins added a commit to ChrisCummins/labm8 that referenced this issue


          configure: Remove --with_cuda option.

b646886

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

[Exported from 65210d58caddc06dd0aa321c4afdc83dfe21de71]

ChrisCummins added a commit to ChrisCummins/labm8 that referenced this issue


          Remove ./configure script.

7108cb8

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

[Exported from 0083dd84a4b694deab2162f8ba9c912e08b755ea]

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          configure: Remove --with_cuda option.

3789ac0

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

[Exported from 65210d58caddc06dd0aa321c4afdc83dfe21de71]

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          Remove ./configure script.

4cf4bcb

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

[Exported from 0083dd84a4b694deab2162f8ba9c912e08b755ea]

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          configure: Remove --with_cuda option.

a1ba4ea

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

[Exported from 65210d58caddc06dd0aa321c4afdc83dfe21de71]

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          Remove ./configure script.

cfd85a7

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

[Exported from 0083dd84a4b694deab2162f8ba9c912e08b755ea]

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          configure: Remove --with_cuda option.

527c71c

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          Remove ./configure script.

87b0894

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          Remove ./configure script

6f93a22

github.com/ChrisCummins/phd/issues/73

ChrisCummins added a commit to ChrisCummins/cldrive that referenced this issue


          configure: Remove --with_cuda option.

eb2b006

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/cldrive that referenced this issue


          Remove ./configure script.

7d695c0

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/cldrive that referenced this issue


          Remove ./configure script

d42dd86

github.com/ChrisCummins/phd/issues/73

ChrisCummins added a commit to ChrisCummins/cldrive that referenced this issue


          configure: Remove --with_cuda option.

23cdbe5

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/cldrive that referenced this issue


          Remove ./configure script.

36e98c3

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/cldrive that referenced this issue


          Remove ./configure script

f64d2f6

github.com/ChrisCummins/phd/issues/73

ChrisCummins added a commit to ChrisCummins/labm8 that referenced this issue


          configure: Remove --with_cuda option.

de17be1

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/labm8 that referenced this issue


          Remove ./configure script.

eb6f6f0

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/labm8 that referenced this issue


          Remove ./configure script

b5d4f73

github.com/ChrisCummins/phd/issues/73

ChrisCummins added a commit to ChrisCummins/ProGraML that referenced this issue


          configure: Remove --with_cuda option.

a2c1fad

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/ProGraML that referenced this issue


          Remove ./configure script.

6471dd0

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/ProGraML that referenced this issue


          Remove ./configure script

0facd5e

github.com/ChrisCummins/phd/issues/73

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          configure: Remove --with_cuda option.

9a6084c

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          Remove ./configure script.

4ab9641

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          Remove ./configure script

7bf07fc

github.com/ChrisCummins/phd/issues/73

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          configure: Remove --with_cuda option.

78ef970

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          Remove ./configure script.

1d28834

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/clgen that referenced this issue


          Remove ./configure script

d779310

github.com/ChrisCummins/phd/issues/73

ChrisCummins added a commit to ChrisCummins/ProGraML that referenced this issue


          configure: Remove --with_cuda option.

df99e2d

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/ProGraML that referenced this issue


          Remove ./configure script.

13b855e

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/ProGraML that referenced this issue


          Remove ./configure script

5a8c919

github.com/ChrisCummins/phd/issues/73

ChrisCummins added a commit to ChrisCummins/labm8 that referenced this issue


          configure: Remove --with_cuda option.

3a10e66

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/labm8 that referenced this issue


          Remove ./configure script.

13a7967

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/labm8 that referenced this issue


          Remove ./configure script

b7ceddd

github.com/ChrisCummins/phd/issues/73

ChrisCummins added a commit to ChrisCummins/ProGraML that referenced this issue


          configure: Remove --with_cuda option.

92ab446

This removes the --with_cuda option from ./configure, forcing the base
tensorflow package to always be used, rather than re-writing it to
tensorflow-gpu. Because of this, I no longer need to generate a
requirements.txt file from tools/requirements, so that can be removed,
simplifying the build process.

For now, I retain the 'with_cuda' field in the config.pbtxt proto, but
hardcode its value to False. Future patches will replace checks on
this field to calls to runtime libraries, such as using
GPUtil.getGPUs() for Python, defined in //third_party/py/gputil.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/ProGraML that referenced this issue


          Remove ./configure script.

9895ba5

This patch finally fixes a long-standing gripe that I have had with my
build toolchain with the requirement to run a ./configure script. By
removing the conditional TensorFlow GPU package, I finally feel ready
to let go of this configuration script and the generated
config.pbtxt. The last remaining piece of functionality ported over
from ./configure is the recording of the workspace source path, which
has been assimilated into build_info.pbtxt.

github.com/ChrisCummins/phd/issues/73

Signed-off-by: format 2020.01.16 <github.com/ChrisCummins/format>

ChrisCummins added a commit to ChrisCummins/ProGraML that referenced this issue


          Remove ./configure script

a125b6e

github.com/ChrisCummins/phd/issues/73

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment