Travis spellcheck and cppunit with valgrind #363
Conversation
|
Ok, so after some debugging to drop later - the mismatch is in aspell shipped dictionaries on travis hosts == more exceptions needed in nut.dict. |
|
Cool, so that worked with modifying the first line of nut.dict with utf8 ;)
Thx for all your work on NUT @jimklimov
Happy to finally see the actual use of all these personal nights of
works!!!!
I owe you some beers ... or vodka ;)
|
|
I did not yet experiment with utf8 header nor other options, so contributor
names are still a problem :(
…--
Typos courtesy of GMail on my Samsung Android
19.01.2017 23:11 пользователь "Arnaud Quette" <[email protected]>
написал:
Cool, so that worked with modifying the first line of nut.dict with utf8 ;)
Thx for all your work on NUT @jimklimov
Happy to finally see the actual use of all these personal nights of
works!!!!
I owe you some beers ... or vodka ;)
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#363 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABj5lJy90aM6vq5Ze9gv14cqlg5DkEFnks5rT9-qgaJpZM4Logbd>
.
|
…allelize, optimize for speed, separate error causes/reports)
jimklimov
force-pushed
the
travis-spellcheck
branch
from
5f849f4 to
21f3a5d
Compare
* spellcheck targets should fail the build not only if the last document failed the test * make sure to use custom spellcheck dictionary from NUT source tree; revise erroring out if automatic aspell finds issues * also set LC_ALL=C for aspell * revised spellcheck failure message to suggest next steps for the human * comment the spellcheck test; use more portable shell test (instead of bracket); do not fail on empty aspell outputs
…and use it from Travis for now * docs/Makefile.am : also report if the build succeeds despite spellcheck errors (due to this setting)
…gging of aspell setup
… targets: spellcheck spellcheck-interactive doc
…mention the recently added option to skip building docs
…ackage description (to fit in 80 cols)
…) and added link to Travis CI dashboard for NUT branches
…te the NUT custom dictionary
…s CI; add some developer names to the dictionary
jimklimov
force-pushed
the
travis-spellcheck
branch
2 times, most recently
from
6d9081e to
9069a0f
Compare
jimklimov
force-pushed
the
travis-spellcheck
branch
from
8237a0d to
12fa862
Compare
jimklimov
changed the title
|
Seems finally fixed for "UTF8" input and dicts (just one Pawel with polish L got missed by aspell and is now seen as Pawe with a skipped next character that is otherwise deemed invalid in a word), so un-WIPing the PR. |
|
@aquette : PS: I am not quite a vodka guy, but I get the drift ;) |
jimklimov
changed the title
There was a problem hiding this comment.
- you can work on such features in nut repos. branch (would make it easier to add commits on my side, such as the below)
- worths to propagate the spellcheck to manpages, in docs/man/Makefile.am:
** create SRC_ALL_PAGES =
$(SRC_CONF_PAGES)
$(SRC_CLIENT_PAGES)
$(SRC_TOOL_PAGES)
$(SRC_CGI_PAGES)
$(SRC_DEV_PAGES)
$(SRC_SERISPELLCHECK_SRCAL_PAGES)
$(SRC_SNMP_PAGES)
$(SRC_USB_LIBUSB_PAGES)
$(SRC_NETXML_PAGES)
$(SRC_POWERMAN_PAGES)
$(SRC_IPMIPSU_PAGES)
$(SRC_MACOSX_PAGES)
$(SRC_LINUX_I2C_PAGES)
** mod EXTRA_DIST to use SRC_ALL_PAGES
** duplicate docs/Makefile.am spellcheck* and use SRC_ALL_PAGES as SPELLCHECK_SRC
| # See also http://aspell.net/man-html/Through-A-Pipe.html | ||
| # TODO: Is "grep -a" or "grep -b" (treat input as ascii/bin) portable enough? | ||
| # Is "egrep == grep -E" always valid? (maybe all a job for configure.ac) | ||
| # TODO: Define a custom language in-repo or during build, that is english |
There was a problem hiding this comment.
is this and the below still valid?
There was a problem hiding this comment.
Custom language - probably not anymore, at least not until we hit a platform with sufficiently different or old aspell.
Concerns of portable egrep and -a/-b flags - these may still be valid (buildbots might show what works where). If the issues do pop up, the flag may be removed at expense of warnings that unexpected binary input was detected - and those might need to be grepped out so as to not pollute logged reports and/or interpretation of aspell's "$OUT".
| @@ -125,17 +125,75 @@ SPELLCHECK_SRC = $(ALL_TXT_SRC) ../README ../INSTALL.nut ../UPGRADING ../NEWS \ | |||
| ../TODO ../scripts/ufw/README ../scripts/augeas/README ../lib/README \ | |||
| ../tools/nut-scanner/README | |||
There was a problem hiding this comment.
add daisychain.txt to ALL_TXT_SRC since it's missing
|
@jimklimov great work! and thx for the thank ;) |
…le and use that for spellchecker
|
Good point on manpages - done. No copy-paste involved ;) |
jimklimov
force-pushed
the
travis-spellcheck
branch
from
8b4490b to
59e6d3e
Compare
jimklimov
force-pushed
the
travis-spellcheck
branch
from
59e6d3e to
9813ef8
Compare
|
For reviewers : so far my fantasy is exhausted as far as "small initial steps" go, and this PR looks solid enough for merging, as a foundation for possible (or inevitable) improvements. I'd like to avoid it growing into a monster like DMF PRs that take months to review, and to begin bringing benefits to subsequent codebase improvements ASAP ;) |
|
@jimklimov thx, you've addressed all the points from my review. Good to go, and as you said, this is already improving the situation, along with being a solid foundation to build upon. |
For issue #338
This enables Travis tests for cppunit (to whatever extent was available in codebase, which was not much), and fixes spellchecking to do its work (and also fixes a number of discovered spelling errors).
Also adds configure capability to find and optionally enable valgrind, but currently does not go beyond that (e.g. enable debug builds and/or run any special checks under valgrind... not sure how/what to test here - start every binary to display help text and see that nothing went wrong? the commented-away code for travis enablement was slightly improved but is still commented away). This piece is ported from DMF branch which does have a target to make a custom build of everything it needed to test that the parser does not leak.
UPDATE: For lack of better ideas at the moment, added a distcheck taryet with valgrind support that runs the cppunit test programs with leak inspection. This approach can be replicated to add other types of 'make check' as they come.