Lyx files converted from 2.1 to 2.2 contain parbeak and latexpar separators that
would not appear if the lyx file was written in 2.2 from scratch. The script
removes latexpar separators and transforms parbreak separators into plain
separators. Then it displays a diff of the resulting pdf files (requires
diffpdf) for manual control. All lyx files shipped with lyx should be updated in
this way. See #10068.
example usage:
cd lib/doc
../development/tools/separator-convert.sh Math.lyx
This is what the polib version does already. If a translation is marked
fuzzy then it is not used, so if the other file contains a non-fuzzy
translation we do not throw away useful information if we overtake it.
If the -o option is given, we may overwrite existing translations with
translations from the other .po file. In this case, we have to delete or set
the fuzzy flag accoring to the updated translation.
he.po in branch contains a number of translations where the translation is
identical with the original text (e.g. for encoding names) which look suspicios
to me. Therefore I did not want to merge these and leave the decision to the
translator. No information is lost, since these "translations" can easily be
recreated from the original texts by copy-paste.
This is useful if you know that the translation of a particular language has
not been updated at all in master, but the stable version has been worked on.
Some characters (like ") are escaped in .po files. These are handled correctly
by polib, but the self written parser did not yet take them into account.
We need to remove the #~ prefix of each line, since it is re-added after
wrapping, and because otherwise the comparison with entries from polib
does not work.
Now it is possible to specify the target directory, and the command line syntax
follows the standard rules: It uses options for optional arguments, and one
positional argument for the single required argument.
Surprisingly I could not find a tool that merges updated translations into a
.po file in a way that
a) merges only missing translations (does not overwrite existing ones)
b) produces a minimal diff so that there is a chance to manually check the result
Therefore I wrote my own. You can use it for a single language
python development/tools/mergepo.py ../lyx-2.1/po fr
or for all languages:
python development/tools/mergepo.py ../lyx-2.1/po
The python code is not the most elegant one, but it works. What is missing is
a command line switch to merge changed translations as well. This is useful
for languages that have not yet received any translation update in the
development branch, only in the stable branch: In this case we know that
translations that are not identical in both branches should be overtaken from
the stabkle one.
Thanks Scott for the idea to modify the document. This seems to work, but I am
not surer whether it is safe in all cases, so better warn if this is used.
gen_lfuns.py does now produce the current file format, and and LFUNs.lyx was
re-created with the updated script.
There is one difference if you compare this version of LFUNs.lyx with the old
version updated by lyx2lyx: All occurences of LyX, TeX etc. in the lfun
descriptions are no longer output as logos. I do consider this as a feature,
since the old version did also output the TeX part of BibTeX as a logo, as well
as places where the names were part of some syntax, e.g. lyx::LyXRC::LyXRCTags.
lyx2lyx did not yet know about /systemlyxdir/ and set \origin to the path
where my git tree lives instead. This path is not usable except on my machine,
so better write something more usable instead.
This is a special command line switch of lyx2lyx, so it does not interfere
with normal usage. I did not try to deduce the systemlyxdir from lyx2lyx to
be on the safe side.
This makes the script usable on windows and speeds it up by an order of
magnitude, since no new process needs to be forked for each layout file.
It also does not conevrt .old files again.
The tqrget now works properly when out of source directory qnd uses the $(PYTHON) variable. Also, the use of the error() function has been fixed in the gen_lfun.py script.
I see lot of errors like:
The following string was expected to be '.cpp' or '.h':
LyX.cpp: In member function 'void lyx::LyX::printError(const lyx::ErrorItem&)':
Warning: the error was not parsed correctly.
The following string was expected to be '.cpp' or '.h':
LyX.cpp: In function 'void lyx::error_handler(int)':
The script does the following:
All .cpp and .h files in the current directory and subdirectories
are checked to see which include statements could be omitted without
causing a build error.
Many of these omissions would not be desired. For example, currently
if you don't include Undo.h in Undo.cpp, there is no error because
Undo.h is included in Cursor.h which is included in Undo.cpp. But
clearly we do want to include Undo.h in Undo.cpp.
See #6305.
Some macros defined by wasysym.sty work only in text mode: They either
produce an error in math mode, or wrong output. These symbols are now marked
as text symbols, so that no \ensuremath is created for LaTeX export if they
appear inside \text{}, and the correct images are created.
development/tools/generate_symbols_images.py now checks for each symbol in
lib/symbols whether its image is listed in lib/Makefile.am and whether it has
a toolbar entry in lib/ui/stdtoolbars.inc.
