Name already in use

Fixes #247.

Django has a runtime dependency on distutils, so adding setuptools as a dependency resolves it.

This has already been resolved upstream [1], but this allows us to keep using the same version of Django for the benchmark.

[1] django/django#14240

The Dask scheduler is largely pure-python-code-bound. There is a discussion about the details here. For this reason, it makes a good real-world workload of a distributed system benchmark (even when the benchmark is running locally on a single core).

Thanks to @TomAugspurger for the suggestion, and @mrocklin for providing the meat of the benchmark itself.

This, in particular, is to get the new pystats support in pyperf
as part of:

  faster-cpython/tools#115

This helps reduce compilation time on multi core machines.

This adds a benchmark of Docutils as an application. I thought a reasonable test load was Docutils' own docs (takes ~4.5-5s on my computer).

I haven't submitted a benchmark before---I don't know the best way of storing the input data, so for speed I copied the documentation into git here (the docs are public domain).

This addresses several of the problems noted in #234.

Notably:

* disallow groups named "all"
* disallow duplicate group names, even if in different included manifests
* disallow benchmarks named "all"
* disallow duplicate benchmark names
* clean up the "list_groups" command
* fix how group resolution falls back to the tags

We weren't properly resetting _resolve when switching between files. So if the default manifest came first in a "combined" manifest, any includes or benchmarks after that would erroneously use the default resolve_default_benchmark().

…232)

The compile and compile_all commands:

- (a) build a fresh Python and `pip` installs `pyperformance` into it.
- (b) for each benchmark, creates a virtual environment and installs
  `pyperformance` (and other things) into that.

If hacking on a checkout of `pyperformance`, you want to make sure that it's
installing from the local checkout and never from PyPI or your changes won't be
in effect.

There are two bugs related to this one:

- pyperformance detects if it's a dev version by looking for an
  `pyperformance.egg-link` file in `site-packages` [1]. If that's not the case,
  pyperformance is installed from PyPI. In step (a), it installs the local
  `pyperformance` in non-editable mode, thus no `.egg-link` file, so when (b)
  happens, `pyperformance` is installed from PyPI.

- The check for the `.egg-link` file itself is broken, but perhaps because
  `toml` changed from a module to a package at one point. It needs to look up a
  directory.

[1] https://github.com/python/pyperformance/blob/main/pyperformance/__init__.py#L32

sqlglot is a pure python sql parser, transpiler, and optimizer

* Support reporting geometric mean by tags

This requires the following changes to pyperf first: psf/pyperf#132

* Ensure `tags` is always a list

* Use property

* Update pyperf

It is useful, especially when creating custom versions of dependencies that are compatible with CPython main, to specify local file paths as dependencies. Unfortunately, that's blocked by a feature that treats any dependency that matches a file path as a requirements file. This makes that check a little more specific which allows for installing wheels.

* Port pprint performance tests

* Benchmark _safe_repr only if it exists

* Rename the benchmark

* Remove a spurious `cmd` variable

* Move constant object to a module level

* Add a description

* Rename benchmark labels

* Make the benchmark top-level

* Return accidentally removed full test names

(And fix a return type)

* Add benchmark for deepcopy

* fix typo

* Update pyperformance/data-files/benchmarks/bm_deepcopy/run_benchmark.py

Co-authored-by: Eric Snow <ericsnowcurrently@gmail.com>

* Update pyperformance/data-files/benchmarks/bm_deepcopy/run_benchmark.py

Co-authored-by: Eric Snow <ericsnowcurrently@gmail.com>

* address review comments

* Update pyperformance/data-files/benchmarks/bm_deepcopy/run_benchmark.py

Co-authored-by: Eric Snow <ericsnowcurrently@gmail.com>

Co-authored-by: Eric Snow <ericsnowcurrently@gmail.com>

Co-authored-by: Guido van Rossum <gvanrossum@gmail.com>

Add a benchmark for testing async workloads, specifically an async tree workload that simulates simpler versions of a typical Instagram endpoint.  (See python/cpython#91121.)

* github: Update PyPy version from 3.6 to [3.7, 3.8, 3.9]

* revert the comment

in manifest group sections

fixes #191