Skip to content

Commit

Permalink
v1.1.0
Browse files Browse the repository at this point in the history
  • Loading branch information
nlgranger committed Dec 26, 2020
1 parent 93bf947 commit 7ee3aa3
Show file tree
Hide file tree
Showing 5 changed files with 19 additions and 10 deletions.
8 changes: 3 additions & 5 deletions .circleci/config.yml
Original file line number Diff line number Diff line change
Expand Up @@ -27,7 +27,7 @@ jobs:
- run:
name: Upload coverage
command: |
tox -e coverage
tox -e coverage -- xml
bash <(curl -s https://codecov.io/bash)
deploy:
parameters:
Expand Down Expand Up @@ -64,17 +64,15 @@ workflows:
requires:
- tests
filters:
branches:
only: master
tags:
only: v*
only: /^v[0-9]+\.[0-9]+\.[0-9]+$/
twine_repository: pypi
twine_password: $PYPI_TOKEN
- deploy:
requires:
- tests
filters:
tags:
only: v*
only: /^v[0-9]+\.[0-9]+.[0-9]+\.dev[0-9]*$/
twine_repository: testpypi
twine_password: $TESTPYPI_TOKEN
12 changes: 10 additions & 2 deletions CHANGELOG.txt
Original file line number Diff line number Diff line change
@@ -1,9 +1,17 @@
# Changelog
All notable changes to this project will be documented in this file.

## [1.1.0] - 2020-12-26
### Added
- usage of shared-memory for process-based prefetching (requires python >= 3.8)

### Removed
- 'sharedmem' prefetching backend
- support for python <= 3.5

## [1.0.0] - 2019-12-27
### Added
- prefteching using shared memory for zero-copy data transfer
- prefetching using shared memory for zero-copy data transfer
- case and switch functions

### Changed
Expand All @@ -17,7 +25,7 @@ All notable changes to this project will be documented in this file.

## [0.9.0] - 2018-09-10
### Added
- improved documention
- improved documentation

### Changed
- changed user-code error reporting (more idiomatic)
Expand Down
3 changes: 3 additions & 0 deletions docs/examples/dataloader.py
Original file line number Diff line number Diff line change
Expand Up @@ -87,6 +87,8 @@ def __init__(
transfers between workers and the main process. That shared
memory is divided into num_worker * prefetch_factor slots.
- timeout is not implemented
- a pool of shared memory with a fixed size (shm_size) is used for
zero-copy buffer transfers from workers.
"""
# sampling/batching
self.dataset = dataset
Expand All @@ -113,6 +115,7 @@ def __len__(self):

def make_sequence(self):
"""Build a sequence that looks like a dataloader when iterated over."""
# shuffling
if self.batch_sampler:
batch_indices = list(self.batch_sampler)
out = seqtools.smap(lambda bi: [self.dataset[i] for i in bi], batch_indices)
Expand Down
2 changes: 1 addition & 1 deletion tests/test_release.py
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,6 @@ def candidate_for_release():
def test_version():
version = pkg_resources.require("seqtools")[0].version

tag = check_output("git describe --exact-match --tags")
tag = check_output("git describe --exact-match --tags").lstrip('v')

assert tag == version, "Package version and commit tag do not match"
4 changes: 2 additions & 2 deletions version.py
Original file line number Diff line number Diff line change
Expand Up @@ -4,15 +4,15 @@


# This line is updated automatically
version = "1.0.0-r18-geb95003"
version = "1.1.0"

# If we are in the repo, the following script will update the version
# number and update this file, otherwise, we are probably in the source
# distribution and the above version number is up-to-date.
thisdir = os.path.dirname(__file__)
try:
description = subprocess.check_output(
"git describe --tags ".split(),
"git describe --tags".split(),
stderr=subprocess.STDOUT,
cwd=os.path.dirname(__file__),
universal_newlines=True)
Expand Down

0 comments on commit 7ee3aa3

Please sign in to comment.