Skip to content

Commit

Permalink
Update translation documentation for Zanata
Browse files Browse the repository at this point in the history
(cherry picked from commit afb8517)

Related: rhbz#1205285
  • Loading branch information
bcl committed Mar 24, 2015
1 parent d734030 commit 133f017
Show file tree
Hide file tree
Showing 3 changed files with 117 additions and 130 deletions.
2 changes: 1 addition & 1 deletion docs/Makefile.am
Expand Up @@ -19,7 +19,7 @@

EXTRA_DIST = boot-options.txt mediacheck.txt \
lvm_sanity_checks.txt rescue-mode api.cfg making-screenshots \
threads.txt gettext.txt transifex.txt
threads.txt gettext.txt translations.txt

CLEANFILES = api *.xml

Expand Down
129 changes: 0 additions & 129 deletions docs/transifex.txt

This file was deleted.

116 changes: 116 additions & 0 deletions docs/translations.txt
@@ -0,0 +1,116 @@
Translations and anaconda Development
24-Feb-2015
by: David Cantrell <dcantrell@redhat.com>
Brian C. Lane <bcl@redhat.com>
-----------------------------------------------------------------------------

Anaconda, as of version 23.1, is using https://fedora.zanata.com for
translations. You will need an account in order to pull translations until bug
https://bugzilla.redhat.com/show_bug.cgi?id=1172618 (anonymous pull requests)
has been resolved.

The zanata-python-client is used by the build scripts, not the full zanata
package.


CLIENT SETUP
------------
Setting up the zanata-python-client on your system for anaconda builds.

1) Install the zanata-python-client package:
yum install zanata-python-client

2) Create a Zanata account at https://fedora.zanata.org
NOTE: This system is linked to FAS, but the 'remember my login' option
only appears to work for about an hour, not 7 days as it claims.

3) Navigate to Dashboard->Settings->Client and click 'Generate New API Key'

4) Copy the contents of the Configuration text box to ~/.config/zanata.ini
and make sure permissions are 0600.

Now zanata is set up on your system. The translation files will only be pulled
when a 'make release' is done. The 'make dist' step will just create a tar
file of the what we have in our repo. The 'make bumpver' step will also push a
new anaconda.pot file to Zanata, so any string changes are pushed to them on a
regular basis.

NOTE: zanata pull is slow. This is why I only added it to the 'make bumpver'
step.

There are some other procedures related to zanata that will have to be done
when we create new branches or when there are translation errors.


MAKING A RELEASE
----------------

git clean -d -x -f
./autogen.sh && ./configure --disable-static \
--enable-introspection --enable-gtk-doc
make bumpver # zanata pull by dependent po-pull target
git commit -a -m "New version." # DO NOT run 'git clean -d -x -f' after
make && make release # signed tag happens after dist now

The process here is mostly the same. I do not recommend that you run
git clean between 'make bumpver' and 'make release'. The reason is you
will have to run 'zanata pull' again and that's slow, plus translations may
have changed between the two steps.

The 'make tag' step now runs after 'make dist' in case dist generation
fails. That way you don't end up with a partially created dist AND a
bad tag you have to delete.

The 'make scratch' and 'make po-empty' targets will now create empty
translation files so that local test builds can be created without needing a
zanata account.


DEALING WITH ERRORS IN *.po FILES
---------------------------------

Translators sometimes introduce errors in the .po files. What we generally
do is try to fix it if it's an obvious typo, or just revert the change and
go back to the old po file. Reverting is harder now since we are not
storing po files in our repo, but in severe cases we can go and fetch the
last build and pull the affected po file from there and use it to revert the
changes.

Here's an example of a po file error that will halt a 'make release':

rm -f af.gmo && /usr/bin/msgfmt -c --statistics -o af.gmo af.po
af.po:7: field `Language-Team' still has initial default value
af.po:1614: number of format specifications in 'msgid' and 'msgstr[1]' does not match
/usr/bin/msgfmt: found 1 fatal error

In this case, I am going to the last known good af.po. To update Zanata,
I do:

cp /path/to/last/known/good/af.po po/af.po
zanata push --push-type target --lang af


CREATING A NEW ANACONDA BRANCH
------------------------------

When we make a new branch, we need to branch the translation files.

On https://fedora.zanata.com go to the project and version to use as a base.
Select 'New Version+' from the hidden drop down menu on the right. Give it a
name that matches the git branch. eg. f23-branch and select the version to copy
from (usually master). Wait for it to finish processing documents.

Create a new git branch:

git checkout master
git clean -xdf
git checkout BRANCH_NAME

At https://fedora.zanata.com select the new version and then select the
'Download Config file' from the hidden dropdown menu next to 'Settings'. This
will download a new zanata.xml file. Replace the zanata.xml file in the root
directory of anaconda project with this new one.

git add -u
git commit -m "New Zanata config file"
git push --set-upstream origin BRANCH_NAME

0 comments on commit 133f017

Please sign in to comment.