* junit test results; unfortunately this seems to conflict with the verbose out
* fix deps cache: two files can change independently
* verbose print obsoleted by gtest in upsream busted
Rationalize horizontal construction of Menu items (TOC,
Bookmarks, Classic file views) for more even padding.
Align "x" close button diagonaly with top right border and title.
Also add forgotten scale_for_dpi to MultiConfirmBox
Improved some stylistic issues.
* Updated Travis CI reference to CircleCI.
* Switched to CircleCI "shield" style badge to fit in better with the other badges.
The debug output from Coveralls was rubbish. `"Build processing error"`
Besides some basics like checking if we were actually sending
valid JSON that gives us absolutely nothing to work with.
I wrote this whole complicated wofkflow-based config file, but except
in the case of a base rebuild it wouldn't really be any faster than
this simple tweak.
```
defaults: &defaults
docker:
- image: houqp/kobase:0.0.5
environment:
EMULATE_READER: 1
# this is for shellcheck 0.4.5 and lower; can be removed for 0.4.6
LC_ALL: en_US.UTF8
version: 2
jobs:
install-and-build:
<<: *defaults
steps:
- checkout
- restore_cache:
keys:
# binary dependencies require {{ arch }} because there are different CPUs in use on the servers
- deps-{{ arch }}-{{ checksum ".ci/install.sh" }}
# need to init some stuff first or git will complain when sticking in base cache
- run: git submodule init base && git submodule update base && pushd base && git submodule init && git submodule update && popd
# we can't use command output directly for cache check so we write it to git-rev-base
- run: pushd base && git_rev_base=$(git describe HEAD) && popd && echo $git_rev_base && echo $git_rev_base >git-rev-base
- restore_cache:
keys:
- build-{{ arch }}-{{ checksum "git-rev-base" }}
- run: echo 'export PATH=${HOME}/bin:${PATH}' >> $BASH_ENV
- run:
name: setup
command: .ci/before_install.sh
- run:
name: install
command: .ci/install.sh
- save_cache:
key: deps-{{ arch }}-{{ checksum ".ci/install.sh" }}
paths:
- "/home/ko/bin"
- "/home/ko/.luarocks"
# compiled luarocks binaries
- "install"
- run:
name: fetch
command: .ci/fetch.sh
- run:
name: check
command: .ci/check.sh
- run:
name: build
command: .ci/build.sh
- save_cache:
key: build-{{ checksum "base/git-rev" }}
paths:
- "/home/ko/.ccache"
- "base"
- persist_to_workspace:
# Must be an absolute path, or relative path from working_directory
root: "./"
# Must be relative path from root
paths:
# front build
- "koreader-emulator-x86_64-linux-gnu/koreader"
test:
<<: *defaults
parallelism: 4
steps:
- checkout
- restore_cache:
keys:
# binary dependencies require {{ arch }} because there are different CPUs in use on the servers
- deps-{{ arch }}-{{ checksum ".ci/install.sh" }}
# need to init some stuff first or git will complain when sticking in base cache
- run: git submodule init base && git submodule update base && pushd base && git submodule init && git submodule update && popd
# we can't use command output directly for cache check so we write it to git-rev-base
- run: pushd base && git_rev_base=$(git describe HEAD) && popd && echo $git_rev_base && echo $git_rev_base >git-rev-base
- restore_cache:
keys:
- build-{{ arch }}-{{ checksum "git-rev-base" }}
- run: echo 'export PATH=${HOME}/bin:${PATH}' >> $BASH_ENV
- attach_workspace:
# Must be absolute path or relative path from working_directory
at: "./"
- run:
name: test
command: .ci/test.sh
docs-and-coverage:
<<: *defaults
steps:
- checkout
- restore_cache:
keys:
# binary dependencies require {{ arch }} because there are different CPUs in use on the servers
- deps-{{ arch }}-{{ checksum ".ci/install.sh" }}
# need to init some stuff first or git will complain when sticking in base cache
- run: git submodule init base && git submodule update base && pushd base && git submodule init && git submodule update && popd
# we can't use command output directly for cache check so we write it to git-rev-base
- run: pushd base && git_rev_base=$(git describe HEAD) && popd && echo $git_rev_base && echo $git_rev_base >git-rev-base
- restore_cache:
keys:
- build-{{ arch }}-{{ checksum "git-rev-base" }}
- run: echo 'export PATH=${HOME}/bin:${PATH}' >> $BASH_ENV
- attach_workspace:
# Must be absolute path or relative path from working_directory
at: "./"
- run:
name: cleanup
command: .ci/after_success.sh
workflows:
version: 2
build-and-test:
jobs:
- install-and-build
- test:
requires:
- install-and-build
- docs-and-coverage:
requires:
- install-and-build
```
* Enable parallelism in .circleci/config.yml
* Add BUSTED_SPEC_FILE to Makefile testfront
* Use it in .ci/test.sh with xargs for test parallelization
NB This is the dumb method of improving test time.
Ideally we want a workflow fan-in/fan-out approach.
This makes Go to, Skim to and TOC page selection work in
scroll mode, and page given to other module more accurate
(previously, we were one action lagging).
This makes button heights similar in all uses of ButtonTable.
It depended on how the ButtonTable was used in each widget
(previously, first and last row may have different sizes than
the others).
buttontable.lua: more even buttons height whether zero_sep or not
framecontainer.lua: added padding_top/bottom/left/right (similar to
what was done for iconbutton)
The following widgets have been adapted for this, with some
additional fixes:
buttondialog.lua
buttondialogtitle.lua: wider title with adequate padding
confirmbox.lua + multiconfirmbox.lua: dismissable via tap outside
inputdialog.lua + multiinputdialog.lua: more even vertical padding between elements
imageviewer.lua
textviewer.lua
datewidget.lua
timewidget.lua
Additionaly: frontlightwidget.lua: fixed width of progress bar that
was exceeding window width since the Size scaling adjustements
Decide baseline vertical position according to font metrics, instead
of the hardcoded 0.7 (in textwidget, which made the text a little
bit up in its box), and 0.75 (in toggleswitch, which made the text a
little bit down in its box). This usually gives a value around 0.72
or 0.73 with our ui fonts, which looks about right.
ReaderFooter: add bottom padding, now that our text goes a few pixels lower
* shellcheck 0.4.5 fix `LC_ALL: en_US.UTF8` (can be removed for shellcheck 0.4.6)
* hush updated luacheck on `reader.lua`; we're not assigning any variables but `= nil` is redundant
I believe this should be `if isKobo()`, or better yet that the entire
block should be moved to `KoboPowerD:init()` because afaik that is the
only platform where the system doesn't provide trustworthy frontlight
information. But to be absolutely sure that I don't break anything (and I
don't want to spend any time on this atm) I'm temporarily excluding only
Android where this behavior is known to be problematic.
See discussion in https://github.com/koreader/koreader/issues/3118#issuecomment-334995879
References #3118 (using keyword "references" because phrases like "possibly fixes"
result in GH autoclose).
"partial" refresh causes a full (without black flash) refresh on
Kindle (which uses REAGL mode for partial refresh). This causes a
full redraw of widgets, which is a bit distracting with some of them:
- dictquicklookup: when showing next definition
- infomessage: when displaying a new one (Wikipedia Save as epub)
Also fix bottom menu, that even when closed, would still register
bottom area as dirty: this would cause top menu navigation to
cause a full partial refresh, only noticable on Kindle.
* Fix crash with keyboard navigation of onHold buttontables
Would crash when encountering a separator or when the number
of buttons in a row changes.
* Reset previous selected item on new buttontable
This gives the page position to these modules even in scroll mode.
Also, in readerrolling: don't query battery/charging status
when crengine does not need it (used only when it shows its top
progress bar).
394be8a (#2855) and 2b3b310 (#3183) introduced side effects (scroll mode
crashing and TOC being reset and rebuild). This reverts parts of them
not yet reverted.
Avoid unnecessary work in ReaderView:onSetViewMode() and
ReaderRolling:updatePos() that would result in TOC being reset
and rebuilt (which can take time on books with huge TOC) during
reader setup.
Technically 4.7 is probably fine for all but debugging symbols, for which a simple version check and workaround could be added. However, since all of the CI tests run 4.8 there is no regression prevention and it would be more future maintenance trouble than it's worth.
Frans de Jonge
poire-z
Robert
Qingping Hou
Fabian Müller-Knapp