| Age | Commit message (Collapse) | Author |
|---|
|
Direct dependencies are significant for 'imply' even if the symbol has
no properties, so they need to be included to get semantically
equivalent output.
Making the direct dependencies clear is helpful in general too, even if
you can usually infer them from the properties they get propagated to.
|
|
The visibility of a symbol in an y-mode choice can only be n or y, so
it's sufficient to check that it's not n (0).
|
|
|
|
It's the top-level Kconfig file that source's depending on $SRCARCH, not
the top-level Makefile.
|
|
|
|
Previously, setting a choice symbol to y would only update
user_selection on the parent choice and not the symbol's own user_value.
Now both are updated.
The point of the old behavior was to remember the m mode selections of a
choice when it was switched back and forth between m and y mode, which
was a feature I thought the C implementation had. On closer inspection,
the C implementation never had that feature, though it might appear like
it if you only make "lucky" changes (if you never select any symbols in
y mode that were n in m mode).
The new behavior is simpler and easier to understand: Symbol.user_value
now always matches the value assigned in a .config file or via
set_value(), provided the value was well-formed. This might avoid some
special-casing in scripts too.
The loss in usability is pretty minimal.
|
|
More experience working with the API convinced me that it's worth it.
Gets rid of ugly conversions in the menuconfig.py and oldconfig.py
examples, and streamlines some things internally as well.
Include two other small fixes as well:
- Make warnings generated by Choice.set_value() match those generated
by Symbol.set_value().
- Get rid of the input stripping in menuconfig.py. It's not like the
interface is usable as-is anyway, and it just complicates the
example.
|
|
Defining a choice symbol in multiple places to add some properties to it
outside the choice seems to actually be done deliberately by MIPS, but
it's almost guaranteed to be an error if the definition(s) outside the
choice have a prompt (and so can be changed by the user there), so warn
for that.
|
|
Never has en affect.
|
|
Mirrors a warning in the C implementation.
|
|
Personal pet peeve. Should add a warning to the C implementation too.
|
|
Arrange by type, which works pretty neatly except for the pesky ranges
check. This also makes us do less work in the common BOOL/TRISTATE case.
Get rid of _check_select_imply_sanity().
|
|
More important stuff up front.
|
|
In a single location. Having multiple definitions with different prompts
is okay.
|
|
Must have a form compatible with the int/hex, like for defaults.
Also refactor a bit and use a single _int_hex_value_is_sane() helper.
Less duplication.
|
|
If a non-constant default is given, we can still check that it has the
right type.
Break out two int/hex value sanity checking functions.
|
|
|
|
Constant as well as undefined symbols lack menu nodes, so it's
sufficient to check 'nodes'.
|
|
Only bool and tristate symbols can select and be selected.
Also add docstrings to the sanity checking functions.
|
|
Mirrors a warning I added to the C implementation.
|
|
- Choices should have type bool or tristate
- Choice values should have a prompt
Also fix indentation mess-up.
|
|
Have separate functions for checking symbols and choices instead of
mixing them up. Easier to read, and avoids some isinstance() checks.
Add some comments too.
|
|
Instead of failing in more cryptic ways later.
Also use 'orig_type' instead of 'type' for the UNKNOWN check. Oversight.
|
|
These are easiest to check after parsing, since a symbol/choice can be
defined in multiple locations:
- Warn if a symbol or choice defined without a type. Also warn for
choice value symbols defined without a type, even if they
automatically get their type from the choice. This feature isn't
well-known and probably not used deliberately.
- Warn if a choice is defined without a prompt
- Warn of a choice default symbol is not contained in the choice
Also move _name_and_loc_str() from the symbol class to the global scope
and generalize it to be able to handle choices.
|
|
Helpful
|
|
Might break U-Boot if they ever upgrade Kconfiglib, because they have a
'select y' in there (which the C tools don't flag since they only look
for quotes), but it's a one-line fix.
|
|
Implicit submenus are only ever created with a symbol as the base.
_has_auto_menu_dep() would indirectly reject anything that wasn't a
symbol anyway, but it led to redundant work.
Also get rid of _check_auto_menu() and inline it into _finalize_tree().
It's easier to read without the recursive call being hidden inside
_check_auto_menu().
Tweak the comments a bit too.
|
|
Extra unconsumed tokens at the end of lines were previously ignored.
Better to flag it.
Also inline _next_token() and _peek_token() into the syntax checking
variants of _next_token(). This code is pretty hot, and it saves a few %
of parsing time.
|
|
|
|
- _saved_line already handles EOF automagically, so no need to
special-case it at the end of help text parsing.
- Check for EOF earlier in _next_line(). Bit silly to do it after line
joining.
|
|
Now points out the error for stuff like 'config' with no symbol name
following it instead of randomly failing later.
1-2% parsing performance hit tops it seems.
|
|
No need to repeat what's already in the README. Just refer to the GitHub
page.
Include some other minor tweaks to the introductory sections.
|
|
No longer needed after 955ea4 ("Support <, > relational operators with
tristates").
|
|
It's also used for the warning added in 0087b40 ("Warn if a symbol with
unsatisfied deps is selected") now.
|
|
|
|
- Match the simpler strchr()y .config parsing done by the C
implementation
- Spell out \w as [a-zA-Z0-9_]. Easier to verify.
- Use ASCII mode for Python 3 to be consistent with Python 2,
where it's already the default. \s no longer matches obscure Unicode
stuff.
This also speeds up regex matching during parsing by about 15% on Python
3, increasing parsing performance by a few %. Looks like there's a tiny
improvement for Python 2 as well.
|
|
Mirrors 9059a3493ef ("kconfig: fix relational operators for bool and
tristate symbols") in the C implementation.
|
|
Use the "U" flag to open() rather than io.open() to avoid a ~14% parsing
performance hit. See comment.
|
|
Mirrors a warning in the C implementation. Make it a bit more
informative and simpler to decode for people who aren't super familiar
with Kconfig.
The warning is printed when/if the symbol is evaluated, e.g. when
writing a .config or C header or accessing Symbol.str/tri_value. It is
not printed if the symbol value has already been calculated and is
cached and up-to-date, which avoids warning spam.
Example:
config Y_SYMBOL_1
def_bool y
config Y_SYMBOL_2
def_bool y
config SELECTED
bool
depends on !Y_SYMBOL_1
config SELECTING_1
def_bool y
select SELECTED
# Skipped in warning, because n
config SELECTING_2
def_bool n
select SELECTED
config SELECTING_3
def_bool y
select SELECTED if Y_SYMBOL_1
depends on Y_SYMBOL_2
# Defined in multiple locations
config SELECTING_3
Generated warning:
warning: SELECTED (defined at Kconfig:7) has unsatisfied direct dependencies (!Y_SYMBOL_1), but is currently being selected by the following symbols:
SELECTING_3 (value: y, defined at Kconfig:20, Kconfig:26), with direct dependencies "y" (value: y) and select condition Y_SYMBOL_1 && Y_SYMBOL_2 (value: y)
SELECTING_1 (value: y, defined at Kconfig:11), with direct dependencies "y" (value: y)
Real-world example from test suite:
warning: NOT_COHERENT_CACHE (defined at arch/powerpc/platforms/Kconfig.cputype:381) has unsatisfied direct dependencies (4xx || PPC_8xx || E200 || PPC_MPC512x || GAMECUBE_COMMON), but is currently being selected by the following symbols:
AMIGAONE (value: y, defined at arch/powerpc/platforms/amigaone/Kconfig:2), with direct dependencies 6xx && BROKEN_ON_SMP (value: y) and select condition 6xx && BROKEN_ON_SMP (value: y)
|
|
Clarify how to fetch help texts and prompts, mention the no-prompt
warning.
|
|
The fancy regex isn't really justified. Much faster with replace() too,
though this is an unlikely hotspot.
Could have used a \g<0> backreference to refer to the entire match
instead of using a capturing group too. Hadn't discovered that.
Add some selftests for escape() and unescape() too.
|
|
Can be handy to check which symbols got set, like merge_config.py does.
|
|
Ended up with just user_value in the end before releasing Kconfiglib 2.
|
|
Same order as for .config files.
|
|
Can be handy to know that the order of the assignments matches the
Kconfig files.
|
|
Shaves a few % more from _parse_properties().
|
|
Speeds things up a bit further.
Rework the unget handling to save the ungotten line directly instead of
using a flag.
Add some help texts to tests/Klocation to make sure the line number is
updated properly for those.
|
|
Bit easier to read.
|
|
Factors out some code and makes the logic a bit more transparent. It's
only used for help text lines.
|
|
Bit clearer.
|
