CHANGES.md

Changes in 2.2.0

New features:

• date search expression can now find dreams based on arbitrary expressions like “yesterday” or “three weeks ago” (requires GNU date or compatible with the -d option).
• tabulate search expression allows easy generation of a table of matching dreams and arbitrary headers you select. See dr help tabulate for details.
• New -t switch to find search expression allows calculating the proportion of dreams matched.

Bugs fixed:

• Message no longer erroneously suggests running help dr instead of dr help.
• -f option to dr cat is now honored as documented.

Changes in 2.1.0

This release of dr adds compatibility for MacOS. Previously I believed it would work, but had not tested it, and as is typical in untested situations, the reality proved more difficult!

Gotchas:

• If you are using GNU sed earlier than 4.2, you will need to update. For that reason, I’m making this a minor version number increment. This is because the -r ERE switch doesn’t work on MacOS and the -E switch that does was only added to GNU sed in 4.2. If you are unable to update, changing this switch back should be a pretty easy patch.

Fixes:

• Don’t break terminal when dr act e is used to abbreviate dr act edit.
• Removed reliance on several GNU extensions I didn’t realize were GNU extensions, particularly to sed.
• Move almost entirely to using extended regular expressions, as those seem to have a better cross-platform definition.
• Handle differences in leading/trailing whitespace better.
• Change shebang to /usr/bin/env bash. This will make it easier to use a version of bash other than the system default if the system version is too old for dr (which needs at least 4.0).

Changes in 2.0.0

This release of dr breaks interface backwards compatibility and removes several low-value features that were not worth maintaining. The format of dream files has not changed. If you have any scripts that call the dr command, you should review this list for breaking changes.

New features

• New -f option to the tagged search to match the entire header line with an ERE rather than using an hregex. This is intended for fields like “Title” which shouldn’t be treated as multiple comma-separated values.
• New date search, which works like dr find date lt 2018-01-01 to find dreams from before January 1, 2018.
• The template for dr new has gotten smaller but can now be customized with a .dream_template file. Review dr help new for details.

Removed features

• stats has been removed due to being buggy and very specific to me. You can easily write external scripts to produce the statistics you want.
• xwinnow has been removed due to lack of convincing use cases. It can be replaced with a combination of OR and AND conditions.
• validate is removed as a synonym of check-validity (this was deprecated in 1.0.4).
• The Python word-count implementation is no longer supported. The C implementation must be built and installed to the system path to get word counts.

Changed behaviors

• The machine-readable format of drwc now separates its fields with tabs rather than spaces, to match the behavior of the standard wc Unix command. section.

Bugfixes

• 1 matches is now properly singular in the results of a find search.
• The tagged search no longer matches the filename as a tag value if the value specified is numeric (e.g., tagged Tags 500 would match dream 500 even if dream 500 didn’t have a tag value of 500).
• get-header now allows an empty search-expression.
• Several fixes for act (some arguments were not being passed forward; terminal could become broken on launching an interactive editor; all dreams were acted on without warning if no input was received).
• winnow now gives a useful error if something that is not a dream filename is piped into it.

Dependency changes

New dependencies:

• python3 is now required, but solely for the regenerate-tags command, so it could reasonably be omitted if you don’t need ctags files. (Python 2 is no longer required or supported.)
• awk is now used fairly extensively. Any POSIX-compliant awk should work.
• drwc (included in this repository) must now be installed to the system path to use the word-count function.

Removed dependencies:

• python2 is no longer required. All Python snippets have either been converted to bash or awk or upgraded to Python 3.
• ed is no longer required due to an increasing number of distros not supplying it in the base package.
• seq, shuf, and tac, which are not POSIX standards, are no longer required. They will be used if available, but dr now includes fallback versions.
• The scripts folder is no longer required in a dreamdir, and dr does not need to access any external files in the dreamdir besides the actual data.

Miscellaneous

• Significant performance improvements to searching, syntax highlighting, and other operations.
• Code cleanup, test coverage improvements, and fixing of shellcheck errors.
• The scripts directory has not been retested or reviewed, and no part of dr itself depends on these anymore. Base your work on example scripts at your own risk.

Changes in 1.0.5

Bugfixes:

• Work around vim's system() getting messed up by a nonstandard umask: just change the permissions with chmod after creation rather than using umask.

New features:

• You can place, e.g., 'dr act dump-headers' at the end of a 'winnow' pipeline to run dump-headers on the results, rather than having to use shell command substitution like 'dr dump-headers $(dr find ...)'.
• Vim tags support, and a generation script and 'regenerate-tags' action to go with it. See the Vim section of the README for details.

Changes in 1.0.4

This release mainly consists of internal cleanup and fixes introduced as a result of static analysis with 'shellcheck' and an extended test suite, but a few user-facing changes have been made as well.

Bugfixes:

• Fix 'random' not shifting away its arguments and preventing any subsequent search expressions from working.
• Don't break when additional search expressions are placed after an expression with an optional argument that is not provided (e.g., 'dr find random tagged Places home').
• 'validate' has been changed to 'check-validity' to resolve an abbreviation conflict with 'version'. 'validate' is retained as a deprecated alias of 'check-validity'; 'v' now works only for 'version'.

New features:

• "headerdex" script to create an index from the headers of all dreams.
• Numerical ranges can now be used as search expressions (1-5, 1-, and -5 are all valid ranges, with the open end being replaced with the first or last dream as appropriate).
• 'cat' and 'dump-headers' will now color their output in a manner similar to the vim syntax highlighting file, if output is to a terminal. Note that this may cause a noticeable slowdown if printing a very large amount of output to a terminal.
• 'stats' now uses the improved Dreamdir word counts rather than the system 'wc' command. This means your word counts may drop noticeably because your headers are no longer being counted as part of your dream text.

Under-the-hood improvements:

• Add use of 'shellcheck' to the pre-commit hook script, and add notes about running these tests before contributing any code.
• Too many others to list -- please see the commit history for details.

Changes in 1.0.3

This release improves and normalizes a number of search functions. Despite all the changes, I think the new syntax in this version is entirely backwards-compatible with 1.0.2.

Bugfixes:

• Fix limit option to search expressions not working at all.
• Don't display '1 dreams' anywhere.

New features:

• Improve 'titleadd.sh' script, rename to 'headeradd.sh' to reflect widening function.
• Add empty search expressions and search expression 'all' to match all dreams.
• Adjustments to online help formatting and content.
• Greatly improve runtime of header-replace, and make it take any search expression rather than just a list of filenames to apply the changes over.
• Change separator used for 'sed' in header-replace so that you can use pipes in your search regex.
• list-headers and header-values now accept search expressions to limit their application to a subset of dreams.

Under-the-hood improvements:

• Change a bunch of variables used in functions to local ones.

Changes in 1.0.2

This is primarily a bugfix release.

Bugfixes:

• Add an ftdetect file so the vim plugin actually loads on .dre files out of the box.
• Remove dependency on a script on my local machine that wasn't included in the distribution.
• The formatting of 'dr find' is no longer messed up if dreams with ID numbers greater than 9999 are included in the output.
• Search expression options -vrsl are now applied in a consistent and documented order.
• Dream files are now pulled with the glob '[0-9][0-9][0-9][0-9][0-9].dre' rather than '*.dre', so other .dre files in the directory that are not dreams (for instance, temporary files that you wanted syntax highlighting to work on) will not confuse the search functions.
• Fix tests several times, and add a pre-commit hook that can be set to run tests before committing.
• Fix RegeneratePlots.sh getting very confused when the scripts/ directory in a dreamdir is a symlink.

New features:

• Add a -f option for 'dr cat' that folds lines at 80 characters, trying to do so at word boundaries (using 'fold -s').
• Print word counts for each individual dream, not just the total (and add an option that prints only the total).

Changes in 1.0.1

Bugfixes:

• Fix a bug where '.' was not correctly handled in search regexes.
• Add a blank line at the end of the headers when creating a new dream; not doing so was an accidental regression.
• Fix a problem where options set by the vim plugin only applied to the first buffer if multiple files were loaded.
• Display the average words/day, not the average words/dream, as the "average" in the statistics table.

New features:

• Add a new script to help with adding titles to dreams that don't have them.
• Implement word count feature in C, and add a front-end in dr.
• Write some tests for the dr script; more to come.

vim:tw=80: