From 8b6cac565676488dae545a30a58bb785584a9d51 Mon Sep 17 00:00:00 2001 From: "Martin A. Brown" Date: Wed, 27 Apr 2016 15:56:11 -0400 Subject: [PATCH] commit of ChangeLog created from git history --- ChangeLog | 207 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 207 insertions(+) create mode 100644 ChangeLog diff --git a/ChangeLog b/ChangeLog new file mode 100644 index 0000000..1b6378a --- /dev/null +++ b/ChangeLog @@ -0,0 +1,207 @@ + +2016-04-21 Martin A. Brown + * bumping version to tldp-0.7.7 + * Debian packaging attempt #1, created build with 'native' source format + which will not be accepted + * add debian/copyright file + * ldptool manpage (sphinx-generated for Debian; statically installed in RPM) + * switch --detail reporting to use predictable DOCTYPE and STATUSTYPE names + +2016-04-09 Martin A. Brown + * bumping version to 0.7.5 + * remove 'random' text from .LDP-source-MD5SUMS + * remove the --builddir if empty after complete run + +2016-04-02 Martin A. Brown + * bumping version to 0.7.2 + * using filesystem age for determining build need will not work; switch + to using content hash (MD5) to determine whether a rebuild is necessary or + not + * create .LDP-source-MD5SUMS in each output directory that lists all of + the hashes of the source files used to create that output directory + * remove testing and references to statfiles() and supporting friends + * add a 'lifecycle' test to the testing suite + * report on running success and failure counts during the run (to allow + interruptability if the user wishes) + +2016-03-28 Martin A. Brown + * bumping version to 0.7.0 + * support better handling of --verbose; --verbose yes, --verbose false + * update and improve documentation in stock configuration file + * provide better feedback on directory existence (or not) rather than + silently doing something unpredicable + +2016-03-27 Martin A. Brown + * bumping version to 0.6.7 + * correct situation where publish() was not propagating errors returned + from the build() function; add test + * add broken example Docbook 4 XML file to test suite + * use unicode_literals in all testing code, too + +2016-03-24 Martin A. Brown + * bumping version to 0.6.2 + * fix all sorts of runtime requirements to build under Ubuntu + and run the full test suite on Travis CI + +2016-03-15 Martin A. Brown + * bumping version to 0.6.0 + * full support for Python3, all unicode-ified and happy + * add test to fall back to iso-8859-1 for SGML docs + * success testing with tox under Python 2.7 and 3.4 + +2016-03-14 Martin A. Brown + * bumping version to 0.5.5 + * use sgmlcheck for Linuxdoc sources + * adjust reporting of discovered documents + * use context to prevent more FD leakage + * begin changes to support Python3; e.g. io.StringIO, absolute_import + unicode changes, lots of codecs.open(), unicode_literals, + +2016-03-11 Martin A. Brown + * handle EPIPE and INT with signal.SIG_DFL + +2016-03-10 Martin A. Brown + * bumping version to 0.5.3 + * create long running tests that exercise more of the code in the likely + way that a user would use the utility + * add testing for Docbook 5 XML + * improve look and consistency for --list (--detail) output + * improve README.rst + +2016-03-09 Martin A. Brown + * remove unused markdown and rst skeleton processors + * pass **kwargs through all processor tools + +2016-03-07 Martin A. Brown + * add support for --builddir, ensure that --builddir is on the same + filesystem as --pubdir + * add new option --publish; can't replace a directory atomically, but + get as close as possible by swapping the newly built output (from + --builddir) with the old one (formerly in --pubdir) + * switch to using 'return os.EX_OK' from functions in driver.py that + can be tested and/or wrapped in sys.exit(function(args)) + * testing improvements for Asciidoc and driver.py + +2016-03-06 Martin A. Brown + * provide user-discoverable support for --doctypes and --statustypes + * correct removal of Docbook4XML generated source document during build + +2016-03-05 Martin A. Brown + * use a simplified technique (arbitrary attributes on function objects) + to generate the DAG used for topological sorting and build order + generation (thanks to Python mailing lists for the idea) + +2016-03-04 Martin A. Brown + * bumping version to 0.4.8 + * add FO generation XSL + * do not set a system default for --sourcedir / --pubdir (user must + specify, somehow) + * DocBook5/DocBook4: process xincludes before validation with xmllint + * add support for AsciiDoc detection and processing + +2016-03-03 Martin A. Brown + * validate all documents (where possible) before processing + * provide support for DocBook 5.0 (XML) + * correct --loglevel handling in driver.py (finally works properly!) + * complete support for --script output + +2016-03-02 Martin A. Brown + * bumping version to 0.4.5 + * fix handling of STEMs which contain a '.' in the name + * review signature identification in each DOCTYPE processor and + validate and reconcile errors with PUBLIC / SYSTEM identifiers + for the SGML and XML declarations + * make sure that build() exits non-zero if ANY build fails + +2016-03-01 Martin A. Brown + * bumping version to 0.4.2 + * support a system configuration file /etc/ldptool + * add entry points and make first full installable build + * allow empty OutputDirectory() object + * begin overhauling the porcelain in driver.py + +2016-02-29 Martin A. Brown + * overhaul generation of inventory object from sources/outputs + * add command-line features and options; actions in particular + * continue improving coverage, at 100% on utils.py + * complete CascadingConfig object creation + +2016-02-26 Martin A. Brown + * generate a DAG for each processor class, so dependencies can + be localized (controlled, abstracted) to each processor class + * use topological sort of the DAG to drive generation of the shellscript, + which leads to massive simplification of the generate() method + * user can specify explicit file to process + * better PDF generation logic (relying on jw) + * provide support for --script outputs (logical equiv. of --dryrun) + * if a document processor is missing prerequisites, gripe to logging + and skip to the next document + * support a SourceDocument named by its directory + * add timing to each processor (some documents take minutes to process, + others just a few seconds; good for users trying to understand which...) + +2016-02-25 Martin A. Brown + * overhaul where and how logging module gets called; driver.py is main + * adding --skip feature; can skip STEM, DOCTYPE or STATUSTYPE + * automatically detect configuration fragments in document processors + with object inspection + +2016-02-23 Martin A. Brown + * add support for --detail (and --verbose) for both source and output docs + * pass args into all driver functions + * get rid of platform.py and references (not necessary any longer) + * fix FD leakage in function execute() and add test case (prevent reversion) + (and start switching to contextlib 'with' usage to avoid in future) + * start generalizing the build process for all doctypes in common.py + * move all generic functionality into BaseDoctype object + * revise fundamental execution approach; generate a shellscript (which can + be executed or simply printed) + * make logging readability improvements: clarity, relevance and alignment + +2016-02-22 Martin A. Brown + * adding ArgumentParser wrapper so can support config file + envars + * all sorts of work for support cascading configuration + * allow each processor to have its own configuration fragment, e.g. + --docbook4xml-xmllint; owned by the Docbook4XML object + * add support for --dump-cfg, --dump-env, --dump-cli, --debug-options + * adding the license text (MIT) and all of that stuff + * creating and fixing the setup.py + +2016-02-19 Martin A. Brown + +2016-02-18 Martin A. Brown + * process and report on documents in case-insensitive stem-sorted order + * add many docstrings for internal usage + * move all source directory scanning logic out of the SourceCollection + object; easier to test and simpler to understand + +2016-02-17 Martin A. Brown + * add logic for testing file age, assuming a fresh checkout of the + source documents; use filesystem age to determine whether or not + a document rebuild is necessary + * initial support for driver.py (eventually, the main user entry point + and inventory.py (for managing the identification and pairing of + source and output documents) + +2016-02-16 Martin A. Brown + * adding tons of testing for document types, edge cases, duplicate + stems, sample valid and broken documents + +2016-02-15 Martin A. Brown + * first processor, Linuxdoc, reaches success + * provide better separation between a SourceCollection and the + individual SourceDocuments; analogously, between OutputDirectory + and OutputCollection + * provide similar dict-like behaviour for SourceCollection and + OutputCollection (which is known to the user as --pubdir) + +2016-02-12 Martin A. Brown + * first processor, Linuxdoc, fleshed out, created (failed) + * generate skeletons for other supported source document formats + * automate detection of source document format; add initial testing tools + +2016-02-11 Martin A. Brown + * core source collection and output directory scanning complete + +2016-02-10 Martin A. Brown + * initial commit and basic beginnings