Make sure we (re)initialize-asdf at some early point: things may have changed after image-dump.

Also, fix Makefile.release on a Mac.

Extract version from git when finalizing the image (requires cl-launch 3.009).
On CCL/windows, work around bug ccl bug 858 in run-program.
Simplify some tests for run-program, fix others.

Extract version from git when finalizing the image (requires cl-launch 3.009).
On CCL/windows, work around bug ccl bug 858 in run-program.
Simplify some tests for run-program, fix others.

Give proper semantics to resolve-absolute-module-name and resolve-module-name.
Make the hello example work properly.
hello now uses the generate functionality to extract version from parent XCVB.
Have log-format do fresh-lines for you.
Make sure builds are only interned once.
Recognize that builds may be invalid because of inherited conflict.

* Introduce the world-grain abstraction as a step towards a farmer.
* A few important cleanups and refactorings.
* better hush the build

* xcvb-master: robustifying the reading of slave results
 (on a suggestion by Attila Lendvai) + doc updates
* xcvb repl shouldn't need you to export SBCL_HOME anymore.

+ shuffling code outside of main.lisp
+ a bit of farming

* start of a forker and farmer
* better error message when you forget to make your dependencies a list
* doc update

Also, clean up the driver.
Preliminary tthsum implementation.
Remove ";;;" from LICENSE.

Don't use asdf:run-shell-command anymore, but have a portablish
xcvb-master:run-program/process-output-stream and derivatives in xcvb-master;
so much cleaner!

Also, lay ground for a non-enforcing build (using POIU?).

Untested. Also, more refactoring.
Moving parse-formals away since it's not so useful and doesn't handle defmethod.

Also, be verbose when removing a forward dependency.

* Correctly sort according to original asdf traverse order
* Use pretty-print for the module form.
* Tweak some internal functions.
* Release with the latest asdf, asdf-dependency-grovel

Requires cl-launch 2.22, a closer2mop update.
(Changes actually by fare)

Moving version to its own file (bump!).
Minor Lisp tweaks (use do-find-symbol in driver, keyword argument to read-first-file-form).

* ensure installation directories exist
* add module forms to every file
* fix nickname registration
* add an eval-when for compile-time functions
* add support for xcvb.mk in the main Makefile
* trivially tested the result

* Incremented version. Since I last did, trivial showstopper bugs were fixed
 (see previous git commits). We sorely need a regression test.

* Moved unused files to a subdirectory unbaked/ and documented them there.

* Updated INTERNALS documentation; sprinkled code with a docstring and a TODO.

* make ASDF superseding work by properly refactoring the dependency-interpreter
* properly recurse on build targets
* record dependencies included in an image
* complete initial image creation
* debug the driver
* properly detect loaded asdf up-to-date status
* error when there are extension forms, for now
* add some verbosity mechanism
* add some trivial command-line interface
* actually use the command-line-arguments library (and debug its packaging)
* complete the mock test example, with a working Makefile!

Note: XCVB 0.300 depends on cl-launch 2.13.3 or later... to be released as 2.14 RSN.

* the Makefile itself is still untested, but XCVB now produces a Makefile that looks like it's work!
* support for building initial image from driver
* update to portablish pathnames
* update asdf frontend
* update mock test case
* debug command-line-arguments packaging
* small documentation update
* remove obsolete files extensions and traverse
* add trivial logging infrastructure
* change the way images are computed -- we can't allow name override anymore since the "name"
  serves as a pattern for how to compute the stuff.

* Doc update: Git repo has moved, etc.
* Beginning of support for BUILDs that actually supersede ASDF systems.
* New cool utility funcallf
* Fix .asd file

* add a BUILD file for xcvb
* adding infrastructure to better track dependencies in general case
* documentation update
* Re-start a Makefile backend

The static backend thingie now compiles and runs to completion. Woot.
Will have to actually make sure its output graph is meaningful,
then write a specific backend to output the graph into Makefile statements.

* Debugged name resolution.
* Debugged fullname inheritance.
* Debugged static backend calls.

At long last, there's the kernel of something that might work.

* created a branch for the v0.1 prototype, documented how to access it.
* rewrote the normalizing and processing of the dependency mini-language.
* separated source vs normalized dependencies in lisp-grain objects.
* compute the latter at the correct time.
* fixed brokenness in the example as these dependencies were normalized (yay!).
* made fullname canonicalization more robust.
* refactored name resolution, fixed broken cases, handle conflicts.
* compiling (if not working) static backend -- to be debugged.
* reworked the misnamed simple-interpreter utility into simple-dispatcher.
* in .asd, comment out files that are borked by the rewrite.
* egoboo: put my name first amongst authors claiming massive rewrite.

* depends-on slots need be initialized before the :after method kicks in.
* can't provide :fullname in parse-module-declaration, or gonna conflict with the specified one.
* detect registry conflicts when trying to resolve names
* stop pretending that packages can isolate xcvb-test from xcvb and then having
 to export symbols (with is not maintainable), import them hackishly (which breaks),
 or have to prefix them all (which sucks). Probably xcvb-static-backends won't survive either.
* small fixes to match previous checkins.

* identify little languages for dealing with dependencies -- dependencies-interpreter.lisp
* hush compiler notes for SBCL
* allow to name pre-images
* fullnames now use leading / to distinguished from unresolved relative names
* now resolving names into actual grains.
* many utilities, including a parser for CL formals
   (over-engineering at work - but how else to be able to portably get user-readable error messages?)
* various small tweaks.