hdl/conda-eda

Conda recipes for FPGA EDA tools for simulation, synthesis, place and route and bitstream generation.

conda-eda's Issues

Additional requirement for building a forked repo with Travis

Apparently the build scripts try to access a repo under conda.anaconda.org with your github username, looking for https://conda.anaconda.org//noarch/repodata.json, which doesn't exist and fails the build. I was able to get past this by signing up to conda.anaconda.org, but maybe if you've already done that under a name other than your github username, you're hosed?

ETOOMANY yosys package

Current there are the following packages distributed in various conda channel:

• for yosys:
  https://anaconda.org/symbiflow/antmicro-yosys
  https://anaconda.org/symbiflow/symbiflow-yosys
  https://anaconda.org/symbiflow/yosys
  https://anaconda.org/symbiflow/yosys-uhdm
  https://anaconda.org/litex-hub/yosys
  https://anaconda.org/litex-hub/symbiflow-yosys
  https://anaconda.org/litex-hub/quicklogic-yosys
  https://anaconda.org/litex-hub/yosys-uhdm

• for plugins:
  https://anaconda.org/litex-hub/yosys-symbiflow-plugins
  https://anaconda.org/litex-hub/symbiflow-yosys-plugins
  https://anaconda.org/litex-hub/quicklogic-yosys-plugins
  https://anaconda.org/symbiflow/symbiflow-yosys-plugins
  https://anaconda.org/symbiflow/yosys-plugins
  https://anaconda.org/symbiflow/uhdm-integration-yosys

SymbiFlow/yosys#96 (comment) suggest that some of those might be obsolete, so we should make sure that this is reflected for the corresponding conda packages (see https://docs.anaconda.com/anacondaorg/user-guide/tasks/work-with-packages/#deleting-a-package).

Move to github actions rather than travis

Travis is broken now....

Discussion: reworking travis stages

The current stage layout is not optimal, as the only purpose it serves is internal organisation.

Since the packages are now split across many repos, it doesn't really matter.

Cons of this setup:

• stages are being built sequentially even when there is no relation between packages
• even if there IS a relation between packages, it doesn't matter, as the second stage can't use the previous stage result unless running on master - otherwise the package is not uploaded to anaconda.

Two feasible solutions I see here:

• have a Stage 1, Stage 2, Stage 3 setup

  • the division might be unclear
  • it makes master green on the first Travis pass
  • doesn't help with PRs, as the

• ignore stages whatsoever

  • easier
  • requires a rerun of Travis on master until all dependencies are built

I am in favor of the first option. The second one can lead to strange results, like one Travis run producing yosys in the latest version and yosys-plugins built against previous yosys.

What are your thoughts on that?

Fix verilator-uhdm build

For some reason it fails on missing GIT_DESCRIBE_TAG.

Fix antmicro-yosys-plugins after renaming yosys-config

yosys-config was renamed in d646059 and it seems to be the root of the antmicro-yosys-plugins failure: https://github.com/hdl/conda-eda/runs/2070486993

I'll add a commit in #90 to skip antmicro-yosys-plugins-linux and antmicro-yosys-complete-linux jobs until this is fixed to make master CI green again.

add klayout package

Upstream: https://github.com/KLayout/klayout
Build recipe: https://github.com/KLayout/klayout#building-without-qt-binding, possibly even without qt with https://github.com/KLayout/klayout/blob/master/build.sh#L93
Dependencies: qt, python, ruby, libcurl, libexpat (but it looks like most of them can be swapped out https://github.com/KLayout/klayout/blob/master/build.sh#L470)

Add package for z3 and boolector for formal?

The Z3 solver for formal verification is quite fast (on some platforms) compared to the default for symbiyosys. On other platforms, boolector is better.

Compilation instructions at https://symbiyosys.readthedocs.io/en/latest/quickstart.html

I'd try to contribute if someone could point me to a clear step-by-step guide. I can also just try to keep attempting pull requests and seeing what the errors from Travis are :)

add openroad package

It would be neat to have openroad available as a conda package.

Maybe doing something similar to http://yowasp.org/ for better cross-platform support?

Build all packages requiring Python additionally with Python 3.8 and 3.9

@mithro (#68 (comment)):

All packages which require Python should build for Python 3.7, Python 3.8, Python 3.9

The reason I was suggesting everything is there is some subtle complexity around the dependencies of Python itself can conflict with dependencies of the other packages.

This can be done in the same way as in the surelog, yosys and symbiyosys recipes from #68.

Instructions in README shouldn't tell you to run travis scripts

nextpnr-nexus: default "main" not tracking most recent build

slang: verible: Use Conda compiler

Currently these two packages are built using a compiler preinstalled in the GitHub Actions Ubuntu image.

They should use a Conda-based compiler.

The "icestorm" package does not contain the "iceprog" executable (Linux)

A user using the conda install ran into an error

FileNotFoundError: [Errno 2] No such file or directory: 'iceprog': 'iceprog'

and I confirmed that iceprog is not in the icestorm package from https://anaconda.org/litex-hub/icestorm.

It is not in the nextpnr-ice40 package either.

Is there another package I need to install to get iceprog?

nextpnr/fpga-interchange: Fix 200t device

The device can't be built due to the OSError: [Errno 12] Cannot allocate memory error: https://github.com/hdl/conda-eda/runs/2504914939?check_suite_focus=true#step:3:4206

It seems to require more than the GH Action's 7GBs of RAM: https://docs.github.com/en/actions/using-github-hosted-runners/about-github-hosted-runners#supported-runners-and-hardware-resources

For now the package will be built without this device.

Updating published sv2v

Firstly, thank you for all your work maintaining releases for HDL tools! I’m sure many appreciate the ease and simplicity this setup provides. I also periodically check in on the results at https://github.com/chipsalliance/sv-tests, which uses the package versions published here.

I've noticed that the version of sv2v under test has recently become stale (~7 weeks old). Although https://anaconda.org/LiteX-Hub/zachjs-sv2v has been updated regularly, it seems that because the other packages over at https://github.com/hdl/conda-misc have had failures, the newer versions of sv2v have not been given the main tag.

• Would it make sense to update the main tag for independent (sets of) packages, even if there are adjacent/unrelated failures?
• Regardless of the above, would it be appropriate to move sv2v from conda-misc to conda-eda? All of the other tools tested by sv-tests are packaged here in conda-eda, while sv2v is packaged in the less-active conda-misc. It seems like sv2v might be more accurately placed here, and would naturally benefit from the greater failure visibility this repo receives.

TRAVIS_OS_NAME must be set before running conda-get.sh

In attempting to do a pseudo-Travis build as in README.md, https://github.com/litex-hub/litex-conda-eda/blob/master/conda-get.sh#L7 requires that TRAVIS_OS_NAME be set, otherwise you'll end up running an OSX version. Which is great if you're on OSX, but not so great otherwise.

some packages don't seems to be compatible with google colab notebooks

It seems that some of the package are not compatible with the runtime environment present in https://colab.research.google.com/.

Installing the following packages with:

!curl -O  https://repo.anaconda.com/miniconda/Miniconda3-py37_4.10.3-Linux-x86_64.sh
!bash Miniconda3-py37_4.10.3-Linux-x86_64.sh -b -f -p /usr/local
!conda install --yes -c LiteX-Hub verible iverilog

produce the following error for verible at runtime:

verible-verilog-lint: /lib/x86_64-linux-gnu/libm.so.6: version `GLIBC_2.29' not found (required by verible-verilog-lint)

while iverilog seems to work fine:

Icarus Verilog version 12.0 (devel) (0_8-5850-g540555fc-dirty)

Installing verible from the older https://anaconda.org/SymbiFlow/repo channel, seems to work fine

I suspect this might be due to #133 updating all packages to be built using a relatively "recent" version of ubuntu, while colab is still an older version (see googlecolab/colabtools#1880).

One solution could simply be to wait for colab to upgrade or provide multiple linux variants using https://docs.conda.io/projects/conda-build/en/latest/resources/variants.html to accomodate for different glibc versions.

Should prjxray-db be architecture agnostic?

If I understand correctly, the database files should not depend on the machine architecture so should probably be marked as working on any?

ci seems to fails consistently on PR branches

See https://github.com/hdl/conda-eda/runs/5078540007?check_suite_focus=true, it fails with the following message:

  Not on master branch, don't execute master-package.sh. Current branch: refs/pull/165/merge
  ERROR: some jobs failed
  Error: Process completed with exit code 1.

Build failure of the Moore package

I skip the moore-linux build, because of the non-trivial fail of the CI. Right now it requires manual building of LLVM.

Sample failure: https://github.com/hdl/conda-eda/runs/5024242928?check_suite_focus=true

I had a super fast attempt to address it here, but of course it's not enough: #162

Documentation build in fomu-workshop fails due to yosys not being found

Since a few days ago, building the Sphinx documentation of fomu-workshop is failing:

https://github.com/im-tomu/fomu-workshop/runs/4751518333?check_suite_focus=true

source env/bin/activate; conda config --system --add envs_dirs /home/runner/work/fomu-workshop/fomu-workshop/docs/env/envs
source env/bin/activate; conda config --system --add pkgs_dirs /home/runner/work/fomu-workshop/fomu-workshop/docs/_download/pkgs
source env/bin/activate; conda env update --name base --file ./environment.yml
Collecting package metadata (repodata.json): ...working... done
Solving environment: ...working... failed

 ResolvePackageNotFound: 
  - yosys

This is the environment file: https://github.com/im-tomu/fomu-workshop/blob/master/docs/environment.yml

Test failure in icarus-osx

The job will be disabled until this is fixed.

Details of a failed run https://github.com/hdl/conda-eda/runs/4782600616

Symbiyosys not showing up in conda packages

https://anaconda.org/litex-hub/repo does not show symbiyosys.

Travis status for litex-hub not found

https://travis-ci.org/litex-hub/litex-conda-eda (from README.md) results in "We couldn't find the repository litex-hub/litex-conda-eda"

nextpnr-ice40/generic: Building on Windows fails due to `undeclared identified`

The error building both nextpnr-ice40 and nextpnr-generic is the same and comes from common/bits.h:

  c:\users\runner\conda-env\conda-bld\nextpnr-generic_1627032814927\work\common\bits.h(54): error C2065: 'v': undeclared identifier [%SRC_DIR%\nextpnr-generic.vcxproj]
  c:\users\runner\conda-env\conda-bld\nextpnr-generic_1627032814927\work\common\bits.h(66): error C2065: 'v': undeclared identifier [%SRC_DIR%\nextpnr-generic.vcxproj]

New prjxray-db is not getting main label?

Looks like the new prjxray-db package (see: https://anaconda.org/LiteX-Hub/prjxray-db/files) is still not getting the main label? In particular, any package built after git hash 057e179 is one that I would've expected to get the label. Any ideas why? I original theorized that it was because of hdl/conda-compilers#13, but I suspect there might be another issue at play?

symbiyosys testing failure: "unexpected TOK_RAND"

Building the package succeeds. Relevant part of the test log:

2021-01-20T18:11:49.8430793Z Error: test failed
2021-01-20T18:11:49.8431818Z SBY 18:11:48 [test/demo3] Writing 'test/demo3/src/demo.v'.
2021-01-20T18:11:49.8432630Z SBY 18:11:48 [test/demo3] engine_0: smtbmc boolector
2021-01-20T18:11:49.8434034Z SBY 18:11:48 [test/demo3] base: starting process "cd test/demo3/src; yosys -ql ../model/design.log ../model/design.ys"
2021-01-20T18:11:49.8435496Z SBY 18:11:48 [test/demo3] base: demo.v:7: ERROR: syntax error, unexpected TOK_RAND
2021-01-20T18:11:49.8436416Z SBY 18:11:48 [test/demo3] base: finished (returncode=1)
2021-01-20T18:11:49.8437202Z SBY 18:11:48 [test/demo3] base: job failed. ERROR.
2021-01-20T18:11:49.8438023Z SBY 18:11:48 [test/demo3] summary: Elapsed clock time [H:MM:SS (secs)]: 0:00:00 (0)
2021-01-20T18:11:49.8439430Z SBY 18:11:48 [test/demo3] summary: Elapsed process time [H:MM:SS (secs)]: 0:00:00 (0)
2021-01-20T18:11:49.8440274Z SBY 18:11:48 [test/demo3] DONE (ERROR, rc=16)
2021-01-20T18:11:49.8441430Z TESTS FAILED: symbiyosys-v0.0_74_g091222b-20210120_172919.tar.bz2

The yosys package (v0.9_5142_g4762cc06) used for testing was built by the same workflow.

Full log with the test script fixed (currently it fails to print log after error on master): https://github.com/litex-hub/litex-conda-eda/runs/1736679449

tree-sitter-verilog: Error while installing the latest tree_sitter v0.2.2 from pip

The last working job building tree-sitter-verilog used tree-sitter 0.2.1. Now I can see that the 0.2.2 installation is attempted but it fails with:

+ /home/runner/work/litex-conda-eda/litex-conda-eda/workdir/conda-env/conda-bld/tree-sitter-verilog_1611736795834/_h_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placeh/bin/python -m pip install --isolated tree_sitter
Collecting tree_sitter
  Downloading tree_sitter-0.2.2.tar.gz (110 kB)
  Installing build dependencies: started
  Installing build dependencies: finished with status 'error'

.
   command: /home/runner/work/litex-conda-eda/litex-conda-eda/workdir/conda-env/conda-bld/tree-sitter-verilog_1611736795834/_h_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placeh/bin/python /home/runner/work/litex-conda-eda/litex-conda-eda/workdir/conda-env/conda-bld/tree-sitter-verilog_1611736795834/_h_env_placehold_placehold_plac
ehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placeh/lib/python3.7/site-packages/pip install --ignore-installed --no-user --prefix /tmp/pip-build-env-5qiezwyi/overlay --no-warn-script-location --no-binary :none: --only-binary :none: -i https://pypi.org/simple -- 'setuptools>=43.0.0' 'wheel>=0.36.2'
       cwd: None
  Complete output (2 lines):
  ERROR: Could not find a version that satisfies the requirement setuptools>=43.0.0
  ERROR: No matching distribution found for setuptools>=43.0.0

It will be skipped by the CI workflow until this is fixed.

Build failures of nextpnr-generic-osx and nextpnr-ice40-osx

The jobs will be disabled until this issue is fixed.

Both jobs share the same problem:

• nextpnr-generic:

  In file included from /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/nextpnr-generic_1615935952290/work/common/idstring.cc:21:
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/nextpnr-generic_1615935952290/work/common/idstring.h:68:16: error: implicit instantiation of undefined template 'std::__1::hash<int>'
          return std::hash<int>()(obj.index);
                 ^
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/nextpnr-generic_1615935952290/_build_env/bin/../include/c++/v1/type_traits:397:50: note: template is declared here
  template <class _Tp> struct _LIBCPP_TEMPLATE_VIS hash;
                                                   ^
  

• nextpnr-ice40:

  In file included from /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/nextpnr-ice40_1615935694227/work/common/idstring.cc:21:
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/nextpnr-ice40_1615935694227/work/common/idstring.h:68:16: error: implicit instantiation of undefined template 'std::__1::hash<int>'
          return std::hash<int>()(obj.index);
                 ^
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/nextpnr-ice40_1615935694227/_build_env/bin/../include/c++/v1/type_traits:397:50: note: template is declared here
  template <class _Tp> struct _LIBCPP_TEMPLATE_VIS hash;
                                                   ^
  

Add package for netgen

Needed by the skywater-pdk to do LVS checks (also needed by OpenRAM).

add open_pdks package

upstream: http://opencircuitdesign.com/open_pdks/
minimal recipe per http://opencircuitdesign.com/open_pdks/install.html:

./configure --enable-sky130-pdk=/path/to/skywater-pdk
make
make install

depends: skywater-pdk #159 https://anaconda.org/LiteX-Hub/magic

Figure out if "Downstream tests" in conda is useful

https://docs.conda.io/projects/conda-build/en/latest/resources/define-metadata.html#downstream-tests

pnr/arachne: Remove `symbiflow-yosys` dependency

Building fails after making such a change (like in #148) but also using yosys instead isn't helpful. make simpletest fails in both cases. It's weird because according to the test and simpletest comments yosys shouldn't be required for the latter: https://github.com/YosysHQ/arachne-pnr/blob/master/Makefile#L101

Arachne shouldn't depend on Yosys and symbiflow-yosys isn't built anymore so this should be fixed.

add xls package

upstream: https://github.com/google/xls
recipe: https://google.github.io/xls/tutorials/hello_xls/#1-installation-and-building
list of produced binaries: https://google.github.io/xls/tools/

Following instructions in README.md tries to install MacOSX binary?

tansell@tansell-glapstation:~/github/litex-hub/litex-conda-eda$ ./.travis/conda-get.sh
+ set -e
+ CONDA_PATH=/home/tansell/conda
+ '[' = windows ']'
./.travis/conda-get.sh: line 7: [: =: unary operator expected
+ '[' = linux ']'
./.travis/conda-get.sh: line 15: [: =: unary operator expected
+ sys_name=MacOSX
+ wget -c https://repo.continuum.io/miniconda/Miniconda3-latest-MacOSX-x86_64.sh
--2020-11-27 17:23:18--  https://repo.continuum.io/miniconda/Miniconda3-latest-MacOSX-x86_64.sh
Resolving repo.continuum.io (repo.continuum.io)... 2606:4700::6812:c94f, 2606:4700::6812:c84f, 104.18.201.79, ...
Connecting to repo.continuum.io (repo.continuum.io)|2606:4700::6812:c94f|:443... connected.
HTTP request sent, awaiting response... 301 Moved Permanently
Location: https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-x86_64.sh [following]
--2020-11-27 17:23:19--  https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-x86_64.sh
Resolving repo.anaconda.com (repo.anaconda.com)... 2606:4700::6810:8303, 2606:4700::6810:8203, 104.16.130.3, ...
Connecting to repo.anaconda.com (repo.anaconda.com)|2606:4700::6810:8303|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 57112343 (54M) [application/x-sh]
Saving to: ‘Miniconda3-latest-MacOSX-x86_64.sh’

Miniconda3-latest-MacOSX-x86_64.sh                   100%[====================================================================================================================>]  54.47M  20.6MB/s    in 2.6s    

2020-11-27 17:23:22 (20.6 MB/s) - ‘Miniconda3-latest-MacOSX-x86_64.sh’ saved [57112343/57112343]

+ chmod a+x Miniconda3-latest-MacOSX-x86_64.sh
+ '[' '!' -d /home/tansell/conda -o '!' -z '' ']'
+ ./Miniconda3-latest-MacOSX-x86_64.sh -p /home/tansell/conda -b -f
PREFIX=/home/tansell/conda
./Miniconda3-latest-MacOSX-x86_64.sh: line 377: md5: command not found
WARNING: md5sum mismatch of tar archive
expected: 6e21e31f666558d485dfdab74c1bc1c8
     got: 
Unpacking payload ...
./Miniconda3-latest-MacOSX-x86_64.sh: line 411: /home/tansell/conda/conda.exe: cannot execute binary file: Exec format error
./Miniconda3-latest-MacOSX-x86_64.sh: line 413: /home/tansell/conda/conda.exe: cannot execute binary file: Exec format error

add simview package

upstream: https://github.com/pieter3d/simview
build system: cmake
dependencies: absl fst surelog ncurses

Packages should have fixed runtime dependencies

Currently packages define runtime dependencies without specifying version (so the latest one is used). We should force a specific version where possible.

An example of such dependency can be seen here https://github.com/hdl/conda-eda/blob/master/pnr/vtr/meta.yaml#L38.
This particular one hit us in the symbiflow-examples repo - newer tbb package does not provide libtbb.so.2 required by VPR, but rather libtbb.so.12.

Travis failed to build yosys on OSX: makefile-conda-config.patch failed

Applying patch: '/private/tmp/conda/syn/yosys/makefile-conda-config.patch'
Trying to apply patch as-is
INFO:conda_build.source:Trying to apply patch as-is
patching file Makefile
Hunk #1 succeeded at 229 (offset 8 lines).
Hunk #2 succeeded at 519 (offset 8 lines).
Hunk #3 succeeded at 901 (offset 8 lines).
patching file tests/arch/ecp5/mux.ys
patching file tests/arch/efinix/mux.ys
Hunk #1 FAILED at 30.
1 out of 1 hunk FAILED -- saving rejects to file tests/arch/efinix/mux.ys.rej
Applying patch: '/private/tmp/conda/syn/yosys/disable-failing-ecp5-mux.patch'
Applying patch: '/private/tmp/conda/syn/yosys/skip-mux16-efinix-test.patch'
Warning: failed to download source.  If building, will try again after downloading recipe dependencies.
Error was: 
Command '['/usr/bin/patch', '-p1', '--ignore-whitespace', '-i', '/private/tmp/conda/syn/yosys/skip-mux16-efinix-test.patch']' returned non-zero exit status 1.

Test failure in verilator-osx

The job will be disabled until the issue is fixed.

Details from https://github.com/hdl/conda-eda/runs/2125128512?check_suite_focus=true#step:3:3385:

  + make -j -f Vhello_world.mk Vhello_world
  In file included from ../sim_main.cpp:1:
  In file included from ./Vhello_world.h:11:
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/verilator_1615926793136/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_pl/share/verilator/include/verilated_heavy.h:808:37: error: implicit instantiation of undefined template 'std::__1::array<unsigned char, 1>'
      using const_iterator = typename Array::const_iterator;
                                      ^
  ./Vhello_world.h:26:33: note: in instantiation of template class 'VlUnpacked<unsigned char, 1>' requested here
      VlUnpacked<CData/*0:0*/, 1> __Vm_traceActivity;
                                  ^
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/verilator_1615926793136/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_pl/bin/../include/c++/v1/__tuple:219:64: note: template is declared here
  template <class _Tp, size_t _Size> struct _LIBCPP_TEMPLATE_VIS array;
                                                                 ^
  In file included from ../sim_main.cpp:1:
  In file included from ./Vhello_world.h:11:
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/verilator_1615926793136/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_pl/share/verilator/include/verilated_heavy.h:812:11: error: implicit instantiation of undefined template 'std::__1::array<unsigned char, 1>'
      Array m_array;  // Contents of the packed array
            ^
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/verilator_1615926793136/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_pl/bin/../include/c++/v1/__tuple:219:64: note: template is declared here
  template <class _Tp, size_t _Size> struct _LIBCPP_TEMPLATE_VIS array;
                                                                 ^
  In file included from Vhello_world__ALL.cpp:3:
  In file included from ./Vhello_world.cpp:5:
  In file included from ./Vhello_world.h:11:
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/verilator_1615926793136/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_pl/share/verilator/include/verilated_heavy.h:808:37: error: implicit instantiation of undefined template 'std::__1::array<unsigned char, 1>'
      using const_iterator = typename Array::const_iterator;
                                      ^
  ./Vhello_world.h:26:33: note: in instantiation of template class 'VlUnpacked<unsigned char, 1>' requested here
      VlUnpacked<CData/*0:0*/, 1> __Vm_traceActivity;
                                  ^
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/verilator_1615926793136/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_pl/bin/../include/c++/v1/__tuple:219:64: note: template is declared here
  template <class _Tp, size_t _Size> struct _LIBCPP_TEMPLATE_VIS array;
                                                                 ^
  In file included from Vhello_world__ALL.cpp:3:
  In file included from ./Vhello_world.cpp:5:
  In file included from ./Vhello_world.h:11:
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/verilator_1615926793136/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_pl/share/verilator/include/verilated_heavy.h:812:11: error: implicit instantiation of undefined template 'std::__1::array<unsigned char, 1>'
      Array m_array;  // Contents of the packed array
            ^
  /Users/runner/work/conda-eda/conda-eda/workdir/conda-env/conda-bld/verilator_1615926793136/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_pl/bin/../include/c++/v1/__tuple:219:64: note: template is declared here
  template <class _Tp, size_t _Size> struct _LIBCPP_TEMPLATE_VIS array;
                                                                 ^
  2 errors generated.
  make: *** [Vhello_world.mk:60: sim_main.o] Error 1
  make: *** Waiting for unfinished jobs....
  In file included from Vhello_world__ALL.cpp:5:
  ./Vhello_world__Slow.cpp:13:101: error: cannot initialize object parameter of type 'const VerilatedModule' with an expression of type 'Vhello_world'
      Vhello_world__Syms* __restrict vlSymsp = __VlSymsp = new Vhello_world__Syms(_vcontextp__, this, name());
                                                                                                      ^~~~
  3 errors generated.

add skywater-pdk package

upstream: https://github.com/google/skywater-pdk
minimal recipe http://opencircuitdesign.com/open_pdks/install.html:

git clone https://github.com/google/skywater-pdk
cd skywater-pdk
git submodule init libraries/sky130_fd_io/latest
git submodule init libraries/sky130_fd_pr/latest
git submodule init libraries/sky130_fd_sc_hd/latest
git submodule init libraries/sky130_fd_sc_hvl/latest
git submodule update
make timing

Add ngspice package

nspice is needed for analog simulation of the spice models in the skywater-pdk (and for OpenRAM characterization).

add cvc package

Upstream: https://github.com/d-m-bailey/cvc
Recipe: typical autoconf/automake https://github.com/d-m-bailey/cvc#installation
Build dependencies: gcc, bison
Extras: GUI and python seems to be split in a separate packages https://github.com/d-m-bailey/cvc/tree/master/src_py so should be easy to stub out

Add ghdl and ghdl-yosys-plugin

In order to use VHDL with the yosys flows, we need ghdl and the ghdl yosys plugin to be part of the environment.

https://github.com/ghdl/ghdl-yosys-plugin

https://github.com/ghdl/ghdl.git

Conda build env adds bunch of compiler flags affecting the resulting binaries

In my local tests I got:

CXXFLAGS=-fvisibility-inlines-hidden -std=c++17 -fmessage-length=0 -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -ffunction-sections -pipe -isystem (...)/conda/include
DEBUG_CXXFLAGS=-fvisibility-inlines-hidden -std=c++17 -fmessage-length=0 -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-all -fno-plt -Og -g -Wall -Wextra -fvar-tracking-assignments -ffunction-sections -pipe -isystem (...)/conda/include
LDFLAGS=-Wl,-O2 -Wl,--sort-common -Wl,--as-needed -Wl,-z,relro -Wl,-z,now -Wl,--disable-new-dtags -Wl,--gc-sections -Wl,-rpath,(...)/conda/lib -Wl,-rpath-link,(...)/conda/lib -L(...)/conda/lib
DEBUG_CFLAGS=-march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-all -fno-plt -Og -g -Wall -Wextra -fvar-tracking-assignments -ffunction-sections -pipe -isystem (...)/conda/include
CPPFLAGS=-DNDEBUG -D_FORTIFY_SOURCE=2 -O2 -isystem (...)/conda/include
DEBUG_CPPFLAGS=-D_DEBUG -D_FORTIFY_SOURCE=2 -Og -isystem (...)/conda/include
CFLAGS=-march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -ffunction-sections -pipe -isystem (...)/conda/include

While -L or -isystem are understandable, there are some flags significantly affecting the resulting binaries e.g. -DNDEBUG or optimizations level options.

The -DNDEBUG flag hits us in chipsalliance/f4pga-examples#127. It disables all the asserts in Yosys, so conda built binaries were never triggering the issue.

(CC @mithro @litghost @mkurc-ant @tmichalak @tgorochowik @acomodi)

conda-build fails when there is no tag available

Most of the recipes starts with a similar boilerplate to derive the version number from the current git tag:

{% set version = '%s_%04i_%s'|format(GIT_DESCRIBE_TAG|replace('v', '') or '0.X.X', GIT_DESCRIBE_NUMBER|int, GIT_DESCRIBE_HASH or 'gUNKNOWN') %}

Which seems to fails if the corresponding git repo has no tags:
https://github.com/hdl/conda-eda/runs/5108676826?check_suite_focus=true

Using |default() instead or or seems to workaround this issue:
8d74975

But we could also be lookin at using GIT_DESCRIBE_TAG (documented at https://docs.conda.io/projects/conda-build/en/latest/user-guide/environment-variables.html#git-environment-variables) to simplify the boilerplate.

Package magic

For the https://github.com/google/skywater-pdk I need https://github.com/RTimothyEdwards/magic and hence it would be good to get it packaged in conda.

magic package doesn't seems to have cairo enabled

As pointed in #27, it doesn't seems that cairo support is enabled.

🍙 echo 'plot svg foo.svg' | conda-env/bin/magic -dnull -noconsole -nowindow 

Magic 8.3 revision 265 - Compiled on Wed Feb  2 15:49:25 UTC 2022.
Starting magic under Tcl interpreter
Using the terminal as the console.
Using NULL graphics device.
Processing system .magicrc file
Using technology "minimum", version 0.0
Cursor not in a window.
"svg" isn't a valid plot option.
The "plot" commands are:
    plot postscript file [layers]    generate PostScript file for what's
                                     underneath the box
    plot pnm file [width [layers]]   generate PNM file for what's
		                     underneath the box
    plot versatec [scale [layers]]    generate plot for selected printer;
		                     scale is multiplying factor for parameter
		                     lambda, for layers underneath the box
    plot parameters [name value]     set or print out plotting parameters
    plot help                        print this help information

/cc @RTimothyEdwards

Need to apt-get install make for local conda build

In README.md, the instructions need to be updated to:

sudo apt-get install coreutils make

as a fresh docker image of ubuntu has no make.

symbiflow-yosys-plugins-osx: A non-BSD `install -D` is used / no Surelog on macOS

This commit broke building on macOS: chipsalliance/yosys-f4pga-plugins@c613a35

There's no install -D in BSD's install. The equivalent switch seems to be simply -d there but I'm not sure.

I've created an issue on the upstream repository: chipsalliance/yosys-f4pga-plugins#74 . Until then this package won't be built on macOS to enable moving other packages onto the main label.