| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* context managers on file r/w + use shutil.copy where it makes sense.
* lxml wants (demands?) that xml files be processed as bytes
* for the phase where we gen the entity files, read as text anyway
* Need to solve a problem where the generated xml is putting the \n in
literally, not evaluating it.
* Fix some examples broken for py3
* Fix more octal constant instances
* Cleanups suggested by PyCharm: staticmethods, two blanks
before class definition, others.
This addresses issues called out in #3300, but is not a complete
solution because the actual doc build step still fails with the
epydoc failures (which aren't directly because of Py3; epydoc
build doesn't work any better on my system with Py3, even with
the forked version with patches).
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
|
|
|
|
|
| |
Write as raw strings:
Two are docstrings that contain a backslash;
the third is an odd expression parenthesized for no good reason
and containing backslashes.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| |
|
|
|
|
| |
for Python 2.7. subproces.Popen() produces bytes without the encoding argument in 3, which is not recognized by 2.7, and doesn't need to be decoded.
|
|\ |
|
| | |
|
| |\
| | |
| | | |
Fix crash when Latin-1 encoded Latex log file is read with Python 3
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Python 3 throws a UnicodeDecodeError when reading a non-utf-8 file
in text mode with default (utf-8) encoding. This happens when T1 fontenc is
used in Latex and a warning in the log file contains e.g. umlauts.
Invalid characters are now replaced.
|
| |/
| |
| |
| |
| |
| | |
overrides= and scanner= have been deprecated for over a decade.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| | |
|
|/
|
| |
https://gcc.gnu.org/onlinedocs/gcc/Preprocessor-Options.html
|
|
|
|
|
|
|
|
|
|
| |
These two have been deprecated since 2010 (about SCons 2.0), commit
935e6985. Methods are removed, setoption for setting them removed, doc is
removed, tests are migrated to test/Removed/*/Old with a sconstest.skip
file so they don't run, and two new tests are added to confirm that
using the functions and setoptions generate exceptions.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|\
| |
| | |
enable experimental persistent caching of vcvars on win32
|
| |
| |
| |
| | |
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
there was still a path through without fully converted data,
where env['ENV'] could contain unicode, so convert the cache hit instead.
This is a little slower but only for Py2 which is going away anyway.
added the vs2019 exec test (vs-14.2-exec) which is just a copy
of 14.0, 14.1 with the required version changed - this had
never been added.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
reading the cache from the json file got us unicode when
running python2, and this broke certain tests when calling
subprocess.Popen: TypeError, because all the environment
wasn't strings, some was unicode.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
msvc env-var cache-read function use 'except IOError'
since 'FileNotFoundError' did not exist on Py2.7/win32.
try to handle error if caceh write fails.
add SCONS_CACHE_MSVC_CONFIG to manpage.
SCONS_CACHE_MSVC_CONFIG can take a pathname to specify
the cache file to use, or it can use the default.
slightly simplified logging setup - logging is stdlib since
py2.3, no need for try block on import.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Setting env var SCONS_CACHE_MSVC_CONFIG enables a filesystem cache
of vcvars results, making them persistent across scons runs. On
test runs (notably our CI system) this helps a lot; in normal usage
where you run one scons invocation at a time instead of many hundreds
in a test run it will make things a little more responsive (esp on
vs2019) but the impact will be much smaller.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|\ \
| | |
| | |
| | |
| | | |
# Conflicts:
# src/CHANGES.txt
|
| |\ \ |
|
| | |\ \
| | | |/
| | |/| |
Add doc for *EMITTER consvars
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Three *EMITTER construction variables had documentation of
"TODO". A fourth, LDMODULEEMITTER, was not mentioned at all.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| | |\ \ |
|
| | | |/ |
|
| | |/ |
|
| | |
| | |
| | |
| | | |
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| | |
| | |
| | |
| | |
| | |
| | | |
"action_timestamps"
As per PR review
|
| |/
|/| |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Adds --install-sandbox to manpage options and to
section on Install builder.
Adds mention of InstallVersionedLib to user guide.
Fixes #3007
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
These options have been deprecated since 2007. They were originally
announced to be disabled in SCons 2.0.0, but that didn't happen.
Left the deprecated-debug-options behavior is in, but the dictionary of
such options is now empty, and there's a new dict of removed options,
and presence in that dict raises an exception.
The four tests that were in test/Deprecated move to a new directory
test/Removed and are simplified just to make sure invocation errors
scons out. (git interprets most of these as remove/add for some reason)
These appear to have been already removed from docs, so no doc impact.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Work on Builder Methods and Variable Substitution man sections - some
wording tweaks, do some formatting markup a little more consistently,
etc.
Tweak the wording for the Command builder, a description shared
between manpage and userguide.
Closes #3030: Environment.xml had many references to detail sections
elsewhere, like "see XXX section below," which don't work well -
this xml is generated into both the manpage and the user guide,
and the locality of saying "below" is incorrect for the latter since
those sections are only in the manpage. Essentially these are changed
to say "see the manpage section XXX". Adding clickable links is of
course also possible, but I recall there was some objection to
cross-document links, so I didn't do that.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|/
|
|
|
|
| |
Apply the patch in the issue, and further tweak the wording.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
| |
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
| |
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
|
|
| |
addresses review comment
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
| |
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
|
|
|
|
|
| |
Also some formatting tweaks, and a couple of typos.
The description of the default tool now lists some
of what that may include.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
|
|
|
|
| |
In places where only the found/not found status is needed, use
the membership operator (in) for checks instead - makes for easier
reading and is considered faster for shorter strings.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Manpage duplicated (triplicated, actually) tons of construction variables
because they were defined in each of the three possible names for the
D compiler. Someone already started a DCommon.xml, this just finishes
off the job of moving them there; the individual compiler-name docs now
do not define any construction vars.
THe ProgramAllAtOnce builder had the same problem, and got the
same treatment.
Note: lots of the cvars don't actually define the term. I don't use D so
I'm not proposing any changes at this time, but it would be simple enough
to merge in some simple wording if someone thinks it's important enough.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
|
|
|
| |
And replaces some written out "construction environment"
and "construction variable" uses with existin entities.
Signed-off-by: Mats Wichmann <mats@linux.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Dictionary is now described as returning a dict only if called
with no arguments; if called with arguments it returns a string
or list of strings (matching the implmenentation).
Note env.Dump() only takes zero arguments or one, it it not
documented as taking the multiple keys, so there's less ambiguity
with it.
Some examples twiddled a little, and in a couple of cases
Dictionary is not used any longer - we might as well just index
into the construction environment since that works.
Fixes #3156
Signed-off-by: Mats Wichmann <mats@linux.com>
|
| |
|
| |
|
|\ |
|
| |\ |
|
| | |\
| | | |
| | | | |
Fix regression in 3.1.0 with removal of DeciderNeedsNode
|
| | | | |
|
| | | |
| | | |
| | | |
| | | | |
code
|
| | | | |
|
| | |/ |
|