Weak points:
- alignof
- js problems
Todo:
- make js work with C++ and mozjs-17
- then mozjs-24
- then mozjs-52
- then mozjs-60
- decrease number of warnings
INIT_OPTION used to initialize union option_value at compile time by
casting the default value to LIST_OF(struct option) *, which is the
type of the first member. On sparc64 and other big-endian systems
where sizeof(int) < sizeof(struct list_head *), this tended to leave
option->value.number as zero, thus messing up OPT_INT and OPT_BOOL
at least. OPT_LONG however tended to work right.
This would be easy to fix with C99 designated initializers,
but doc/hacking.txt says ELinks must be kept C89 compatible.
Another solution would be to make register_options() read the
value from option->value.tree (the first member), cast it back
to the right type, and write it to the appropriate member;
but that would still require somewhat dubious conversions
between integers, data pointers, and function pointers.
So here's a rather more invasive solution. Add struct option_init,
which is somewhat similar to struct option but has non-overlapping
members for different types of values, to ensure nothing is lost
in compile-time conversions. Move unsigned char *path from struct
option_info to struct option_init, and replace struct option_info
with a union that contains struct option_init and struct option.
Now, this union can be initialized with no portability problems,
and register_options() then moves the values from struct option_init
to their final places in struct option.
In my x86 ELinks build with plenty of options configured in, this
change bloated the text section by 340 bytes but compressed the data
section by 2784 bytes, presumably because union option_info is a
pointer smaller than struct option_info was.
(cherry picked from elinks-0.12 commit e5f6592ee2)
Conflicts:
src/protocol/fsp/fsp.c: All options had been removed in 0.13.GIT.
src/protocol/smb/smb2.c: Ditto.
If globhist_simple_search ran out of memory in stracpy(search_url), it
could leave gh_last_searched_title pointing to freed memory and cause
a crash in the next call. Fix by not freeing gh_last_searched_title.
It is then possible to have gh_last_searched_title and
gh_last_searched_url pointing to strings from different searches;
but that was already possible if stracpy(search_title) failed.
Because this bug occurs only in out-of-memory situations and I don't
think ELinks in general has been properly tested in those, the fix is
perhaps not worth mentioning in NEWS and backporting to elinks-0.11.
Documentation strings of most options used to contain a "\n" at the
end of each source line. When the option manager displayed these
strings, it treated each "\n" as a hard newline. On 80x24 terminals
however, the option description window has only 60 columes available
for the text (with the default setup.h), and the hard newlines were
further apart, so the option manager wrapped the text a second time,
resulting in rather ugly output where long lones are interleaved with
short ones. This could also cause the text to take up too much
vertical space and not fit in the window.
Replace most of those hard newlines with spaces so that the option
manager (or perhaps BFU) will take care of the wrapping. At the same
time, rewrap the strings in source code so that the source lines are
at most 79 columns wide.
In some options though, there is a list of possible values and their
meanings. In those lists, if the description of one value does not
fit in one line, then continuation lines should be indented. The
option manager and BFU are not currently able to do that. So, keep
the hard newlines in those lists, but rewrap them to 60 columns so
that they are less likely to require further wrapping at runtime.
straconcat reads the args with va_arg(ap, const unsigned char *),
and the NULL macro may have the wrong type (e.g. int).
Many places pass string literals of type char * to straconcat. This
is in principle also a violation, but I'm ignoring it for now because
if it becomes a problem with some C implementation, then so will the
use of unsigned char * with printf "%s", which is so widespread in
ELinks that I'm not going to try fixing it now.
The previous code just printed time_t directly with "%ld". Now it
instead first casts to time_print_T (currently long) and then formats
with TIME_PRINT_FORMAT (currently "ld"). So the varargs will now
always match with the format string, even if time_t is longer than
long. This still doesn't correctly format time_t values larger than
LONG_MAX, though. But now it is at least easier to find some of the
places that need to be changed to support that.
I located these time_t-to-string conversions by searching for
str_to_time_t, expires, and last_visit. There are still more places
that assume every interesting time_t value fits either in 32 bits or
in a long, e.g. in the cookie editor and in the ECMAScript interface.
Inspired by bug 6.
With regular comments in the definition of the structure itself,
and with xgettext:c-format comments in constants of that type,
if xgettext would otherwise guess wrong; so that translators
will know they'll have to double any percent signs they add.
I didn't regenerate PO files, though.
Introduce a new list, global_history_reap_list, and a new routine,
reap_deleted_globhist_items, which deletes unused items from that
list. In done_global_history_item, do not free the item; just move it
to that list. Call global_history_reap_list in add_global_history_item
and free_global_history.
This changes the init target to be idempotent: most importantly it will now
never overwrite a Makefile if it exists. Additionally 'make init' will
generate the .vimrc files. Yay, no more stupid 'added fairies' commits! ;)
the same accelerator key to multiple buttons in a dialog box or
to multiple items in a menu. ELinks already has some support for
this but it requires the translator to run ELinks and manually
scan through all menus and dialogs. The attached changes make it
possible to quickly detect and list any conflicts, including ones
that can only occur on operating systems or configurations that
the translator is not currently using.
The changes have no immediate effect on the elinks executable or
the MO files. PO files become larger, however.
The scheme works like this:
- Like before, accelerator keys in translatable strings are
tagged with the tilde (~) character.
- Whenever a C source file defines an accelerator key, it must
assign one or more named "contexts" to it. The translations in
the PO files inherit these contexts. If multiple strings use
the same accelerator (case insensitive) in the same context,
that's a conflict and can be detected automatically.
- The contexts are defined with "gettext_accelerator_context"
comments in source files. These comments delimit regions where
all translatable strings containing tildes are given the same
contexts. There must be one special comment at the top of the
region; it lists the contexts assigned to that region. The
region automatically ends at the end of the function (found
with regexp /^\}/), but it can also be closed explicitly with
another special comment. The comments are formatted like this:
/* [gettext_accelerator_context(foo, bar, baz)]
begins a region that uses the contexts "foo", "bar", and "baz".
The comma is the delimiter; whitespace is optional.
[gettext_accelerator_context()]
ends the region. */
The scripts don't currently check whether this syntax occurs
inside or outside comments.
- The names of contexts consist of C identifiers delimited with
periods. I typically used the name of a function that sets
up a dialog, or the name of an array where the items of a
menu are listed. There is a special feature for static
functions: if the name begins with a period, then the period
will be replaced with the name of the source file and a colon.
- If a menu is programmatically generated from multiple parts,
of which some are never used together, so that it is safe to
use the same accelerators in them, then it is necessary to
define multiple contexts for the same menu. link_menu() in
src/viewer/text/link.c is the most complex example of this.
- During make update-po:
- A Perl script (po/gather-accelerator-contexts.pl) reads
po/elinks.pot, scans the source files listed in it for
"gettext_accelerator_context" comments, and rewrites
po/elinks.pot with "accelerator_context" comments that
indicate the contexts of each msgid: the union of all
contexts of all of its uses in the source files. It also
removes any "gettext_accelerator_context" comments that
xgettext --add-comments has copied to elinks.pot.
- If po/gather-accelerator-contexts.pl does not find any
contexts for some use of an msgid that seems to contain an
accelerator (because it contains a tilde), it warns. If the
tilde refers to e.g. "~/.elinks" and does not actually mark
an accelerator, the warning can be silenced by specifying the
special context "IGNORE", which the script otherwise ignores.
- msgmerge copies the "accelerator_context" comments from
po/elinks.pot to po/*.po. Translators do not edit those
comments.
- During make check-po:
- Another Perl script (po/check-accelerator-contexts.pl) reads
po/*.po and keeps track of which accelerators have been bound
in each context. It warns about any conflicts it finds.
This script does not access the C source files; thus it does
not matter if the line numbers in "#:" lines are out of date.
This implementation is not perfect and I am not proposing to
add it to the main source tree at this time. Specifically:
- It introduces compile-time dependencies on Perl and Locale::PO.
There should be a configure-time or compile-time check so that
the new features are skipped if the prerequisites are missing.
- When the scripts include msgstr strings in warnings, they
should transcode them from the charset of the PO file to the
one specified by the user's locale.
- It is not adequately documented (well, except perhaps here).
- po/check-accelerator-contexts.pl reports the same conflict
multiple times if it occurs in multiple contexts.
- The warning messages should include line numbers, so that users
of Emacs could conveniently edit the conflicting part of the PO
file. This is not feasible with the current version of
Locale::PO.
- Locale::PO does not understand #~ lines and spews warnings
about them. There is an ugly hack to hide these warnings.
- Jonas Fonseca suggested the script could propose accelerators
that are still available. This has not been implemented.
There are three files attached:
- po/gather-accelerator-contexts.pl: Augments elinks.pot with
context information.
- po/check-accelerator-contexts.pl: Checks conflicts.
- accelerator-contexts.diff: Makes po/Makefile run the scripts,
and adds special comments to source files.
Ditch the building of an archive (.a) in favour of linking all objects in a
directory into a lib.o file. This makes it easy to link in subdirectories
and more importantly keeps the build logic in the local subdirectories.
Note: after updating you will have to rm **/*.a if you do not make clean
before updating.
All objects defining $(OBJS) will get them and *.a deleted during
make clean.
The all, clean and install rules now implicitly imply their -recursive
counterparts - those will just do nothing in case of $(SUBDIRS) not
defined, so that's ok.
The root makefile is converted as well as some leaf Makefiles. This
also brings in the required infrastructure and adjusts configure.in
appropriately.
I converted only makefiles containing no configurable stuff, since
that'll require more consideration yet.
