3 years agogetting rid of source packages, v7 topic/nodist
Michael D Labriola [Sat, 6 Dec 2014 03:57:35 +0000]
getting rid of source packages, v7

Cleaned up README, removed dead code, updated comments.  Initial source
package removal implementation might just be done...  ;-)

3 years agogetting rid of source packages, v6
Michael D Labriola [Thu, 4 Dec 2014 03:44:04 +0000]
getting rid of source packages, v6

I can successfully build foo-3.0.0-1.i686.brp like this:

LDFLAGS=-L/home/mike/staging/lib ./srp -v \
  --build=../examples/foo-3.0/srp-example.notes --src=../examples/foo-3.0

LDFLAGS=-L/home/mike/staging/lib ./srp -v \
  --build=../examples/foo-3.0/srp-example.notes --src=../examples/foo-3.0 \
  --intree

LDFLAGS=-L/home/mike/staging/lib ./srp -v \
  --build=../examples/foo-3.0/srp-example.notes \
  --src=../examples/foo-3.0/srp-example-3.0.0.tar.bz2

FIXME: There's still a bunch of comment clean up do do...

3 years agogetting rid of source packages, v5
Michael D Labriola [Tue, 2 Dec 2014 03:29:20 +0000]
getting rid of source packages, v5

I can successfully build foo-3.0.0-1.i686.brp like this:

LDFLAGS=-L/home/mike/staging/lib ./srp -v \
  --build=../examples/foo-3.0/srp-example.notes --src=../examples/foo-3.0

FIXME: I haven't tested building intree or building from a source tarball
       yet...

3 years agogetting rid of source packages, v4
Michael D Labriola [Mon, 1 Dec 2014 03:04:19 +0000]
getting rid of source packages, v4

Expanded argument list for notes constructor, fixed ripple up through
cli.do_build.

Moved file checks into notes.check method.

3 years agogetting rid of source packages, v3
Michael D Labriola [Sun, 30 Nov 2014 03:02:20 +0000]
getting rid of source packages, v3

Reworked cli to add --src, --extra, --intree, and --options.

FIXME: not done.  still need to actually check for required flags before
       calling do_build... which is commented out and not implemented yet.
       ;-)

FIXME: left FIXMEs involving moving required file checking from
       core.build_func into notes constructor.

3 years agogetting rid of source packages, v2
Michael D Labriola [Mon, 24 Nov 2014 02:41:32 +0000]
getting rid of source packages, v2

Well, that broke things nicely... definitely tons to do here.  For
starters, am I still allowing source_filename.tar.bz2?  Or is everything
going to require a previously extracted source dir... that sounds silly.

3 years agogetting rid of source packages, v1
Michael D Labriola [Wed, 10 Sep 2014 01:03:42 +0000]
getting rid of source packages, v1

3 years agoupdated referenced year
Michael D Labriola [Fri, 7 Mar 2014 02:35:33 +0000]
updated referenced year

4 years agoMerge branch 'master' of labriola.dyndns.org:/scrap/projects/srp3
Michael D Labriola [Wed, 30 Oct 2013 13:55:51 +0000]
Merge branch 'master' of labriola.dyndns.org:/scrap/projects/srp3

4 years agoautotools: build and install example files
Michael D Labriola [Sat, 26 Oct 2013 18:11:24 +0000]
autotools: build and install example files

This adds an all-local rule to do a 'make dist' in examples and also adds
an install-data-local rule to install the example files and install time.

4 years agoexamples: added --disable-srp-dist flag
Michael D Labriola [Sat, 26 Oct 2013 17:09:56 +0000]
examples: added --disable-srp-dist flag

By default, the example package now doesn't build or install anything.  The
--disable-srp-dist flag actually causes the package to build and install
things.

This is the 1st part of fixing the nested autotools behavior the way we
want.  Now doing a make install from srp's toplevel doesn't build/install
the example programs.

Part 2 is to make srp's install target install the entire example source
tree somewhere.  Probably along with a dist tarball?

4 years agoautotools: added examples subpackage, v1
Michael D Labriola [Sat, 26 Oct 2013 16:37:22 +0000]
autotools: added examples subpackage, v1

This works for now, but we don't want a 1 to 1 mapping of make targets
(i.e., make install shouldn't install example's binaries, it should install
the example source tree).

4 years agoautotools: misc cleanup
Michael D Labriola [Sat, 26 Oct 2013 16:26:47 +0000]
autotools: misc cleanup

Ran autoreconf --force, added some missing files to dist, tidied things up
a bit.

4 years agodb: started actually coding, v5
Michael D Labriola [Fri, 13 Sep 2013 23:52:44 +0000]
db: started actually coding, v5

Only commit db when needed.  I was half asleep when I committed v4... as I
am now.  ;-)

4 years agodb: started actually coding, v4
Michael D Labriola [Fri, 13 Sep 2013 02:18:41 +0000]
db: started actually coding, v4

Ok, we now actually commit the db when we exit.  Kind of important, huh?

4 years agonotes: make less dynamic, v8
Michael D Labriola [Wed, 11 Sep 2013 22:00:27 +0000]
notes: make less dynamic, v8

Tracing through do_install...  Package is being installed, but db is not
being created...

4 years agonotes: make less dynamic, v7
Michael D Labriola [Sun, 8 Sep 2013 19:53:42 +0000]
notes: make less dynamic, v7

Added code to do_build and do_install to detect if we need to instantiate
a feature section for the notes_file.  As such, this also removes the deps
feature's build_prep method (and associated build entry in its
feature_struct) which just instantiated the notes_deps section class.

4 years agonotes: make less dynamic, v6
Michael D Labriola [Sun, 8 Sep 2013 19:40:04 +0000]
notes: make less dynamic, v6

Moved the deps and perms notes section classes from notes module into their
respective feature modules.  Seems to work ok.

4 years agonotes: make less dynamic, v5
Michael D Labriola [Sun, 8 Sep 2013 19:16:38 +0000]
notes: make less dynamic, v5

Fixed some breakage in create and build processing, moved the notes brp
section population to its constructor, and plumbed the brp.built_from_sha
stuff.

Unless I find more things broken down the road, it looks like we're done
fixing ripple through create and build.  Fingers crossed.  Still need to
run through installation.

4 years agonotes: make less dynamic, v4
Michael D Labriola [Sun, 8 Sep 2013 18:50:07 +0000]
notes: make less dynamic, v4

Still find-n-fixing ripple.  Still tracing through do_build, with lots of
side tracks.  Taking the scenic route, we'll say.  ;-)

4 years agonotes: make less dynamic, v3
Michael D Labriola [Fri, 6 Sep 2013 01:40:34 +0000]
notes: make less dynamic, v3

Still find-n-fixing ripple.  Just started do_build, but it's time for bed.

4 years agonotes: make less dynamic, v2
Michael D Labriola [Fri, 6 Sep 2013 00:40:33 +0000]
notes: make less dynamic, v2

Finished initial rewrite of notes_file.  Now find-n-fixing all the ripple.
So far, I've gone through do_create and things look good.

4 years agonotes: make less dynamic
Michael D Labriola [Thu, 5 Sep 2013 01:58:27 +0000]
notes: make less dynamic

This is only half baked, but I need to go to bed.

4 years agodb: started actually coding, v3
Michael D Labriola [Wed, 4 Sep 2013 14:17:18 +0000]
db: started actually coding, v3

Committing code from a few nights ago.  This is where I realized that NOTES
instances can't be pickled... which I think points out some of the
downsides of having the notes implementation be so dynamic.

I would like the db.installed_package constructor to generate a sha of the
notes and manifest objects, and I'd like to do that consistently across the
codebase by feeding the hashlib serialized data from pickle.dumps().  I had
actually been committing the notes instances into a file, and then reading
that into the hashlib, but that's just too complicated (and silly, I
think).

4 years agoupdate TODO
Michael D Labriola [Sat, 31 Aug 2013 17:09:44 +0000]
update TODO

4 years agodb: started actually coding, v2
Michael D Labriola [Sat, 31 Aug 2013 17:09:33 +0000]
db: started actually coding, v2

Fixed exception handling in load

4 years agodb: started actually coding
Michael D Labriola [Thu, 29 Aug 2013 02:19:22 +0000]
db: started actually coding

Ok, at this point, I have a load and commit method that unpickle and pickle
__db from and to disk.  And a class for handling the installed package
instances.

4 years agodb: more rambling
Michael D Labriola [Thu, 29 Aug 2013 01:25:29 +0000]
db: more rambling

4 years agodb: started rambling
Michael D Labriola [Thu, 22 Aug 2013 00:20:55 +0000]
db: started rambling

4 years agofeatures: size, strip_debug, strip_docs now working at install
Michael D Labriola [Sun, 28 Jul 2013 13:09:57 +0000]
features: size, strip_debug, strip_docs now working at install

4 years agoautotools: bit o cleanup
Michael D Labriola [Sun, 28 Jul 2013 13:08:11 +0000]
autotools: bit o cleanup

4 years agoupdated some docs v3.0.0-alpha2
Michael D Labriola [Thu, 25 Jul 2013 01:15:40 +0000]
updated some docs

4 years agoautotools: getting ready for 3.0.0-alpha2 topic/iter_stages
Michael D Labriola [Wed, 24 Jul 2013 02:16:10 +0000]
autotools: getting ready for 3.0.0-alpha2

4 years agocli: added --features to display stage_map
Michael D Labriola [Wed, 24 Jul 2013 01:53:47 +0000]
cli: added --features to display stage_map

4 years agofeatures,blob,cli: plumb system to use new blob objects
Michael D Labriola [Tue, 23 Jul 2013 21:25:14 +0000]
features,blob,cli: plumb system to use new blob objects

4 years agosrp: only tweak sys.path if running from dev tree
Michael D Labriola [Tue, 23 Jul 2013 19:15:51 +0000]
srp: only tweak sys.path if running from dev tree

4 years agoblob: initial commit
Michael D Labriola [Tue, 23 Jul 2013 19:00:31 +0000]
blob: initial commit

This new module defines methods and a class for dealing with our new BLOB
archives.  It includes a C extension that handles actual file extraction,
although it could use a bit more work.

4 years agoautotools: getting things prepped
Michael D Labriola [Tue, 23 Jul 2013 18:37:55 +0000]
autotools: getting things prepped

4 years agotime to seriously compare TarFile vs BLOB vs BLOB w/ C calls, v3
Michael D Labriola [Tue, 23 Jul 2013 03:02:30 +0000]
time to seriously compare TarFile vs BLOB vs BLOB w/ C calls, v3

Added some results

4 years agotime to seriously compare TarFile vs BLOB vs BLOB w/ C calls, v2
Michael D Labriola [Tue, 23 Jul 2013 01:04:07 +0000]
time to seriously compare TarFile vs BLOB vs BLOB w/ C calls, v2

4 years agotime to seriously compare TarFile vs BLOB vs BLOB w/ C calls
Michael D Labriola [Mon, 22 Jul 2013 23:46:34 +0000]
time to seriously compare TarFile vs BLOB vs BLOB w/ C calls

4 years agogitignore: added built files from examples
Michael D Labriola [Mon, 22 Jul 2013 15:15:35 +0000]
gitignore: added built files from examples

4 years agotesting initial blob implementation, v2
Michael D Labriola [Mon, 22 Jul 2013 01:38:20 +0000]
testing initial blob implementation, v2

Added a few more notes

4 years agotesting initial blob implementation
Michael D Labriola [Mon, 22 Jul 2013 01:20:40 +0000]
testing initial blob implementation

This new test script defines a new blob class for exctracting from my new
BLOB file format.  Early testing shows it's a hair slower than the old
tarfile test script, but uses more CPU.  I'm hoping this means it is indeed
less IO bound, which should translate into better multiproc performance.

Still need to test a few things, but assuming this ends up even being on
par with the tarfile speeds, I think I'll try implementing the guts of it
in a C extension module.

4 years agofeatures: started plumbing for iter stages, v10
Michael D Labriola [Fri, 19 Jul 2013 01:59:27 +0000]
features: started plumbing for iter stages, v10

Fixed the TarInfo problem from v9 by changing to a map of
{tarinfo: TarInfo, checksum: b'asdfasdfasdf'}

4 years agofeatures: started plumbing for iter stages, v9
Michael D Labriola [Fri, 19 Jul 2013 00:28:00 +0000]
features: started plumbing for iter stages, v9

Finally added checsum code to the checksum feature and... it doesn't work.
I thought I tested being able to add new data to TarInfo objects a while
back, but I must have been either using Python2 or I just flat out
misremembered...  TarInfo objects are defined using __slots__, so they
don't have a dict, so I can't do this...

4 years agofeatures: started plumbing for iter stages, v8
Michael D Labriola [Thu, 18 Jul 2013 21:38:18 +0000]
features: started plumbing for iter stages, v8

Removed the silly work['manifest'] stuff and added code to actually write
NOTES and FILES to the srp db directory

4 years agofeatures: started plumbing for iter stages, v7
Michael D Labriola [Thu, 18 Jul 2013 21:25:44 +0000]
features: started plumbing for iter stages, v7

Still working on plumbing the install stages...

4 years agofeatures: started plumbing for iter stages, v6
Michael D Labriola [Thu, 18 Jul 2013 00:45:26 +0000]
features: started plumbing for iter stages, v6

Started reworking the install and install_iter functions.  Got tired while
sorting out what to put in the work map to pass back and forth the feature
functions.  Nighty Night...

4 years agofeatures: started plumbing for iter stages, v5
Michael D Labriola [Wed, 17 Jul 2013 00:00:32 +0000]
features: started plumbing for iter stages, v5

Fixed the pickling problem mentioned in v4

4 years agoMerge branch 'master' into topic/iter_stages
Michael D Labriola [Tue, 16 Jul 2013 23:45:07 +0000]
Merge branch 'master' into topic/iter_stages

* master:
  examples: add more deps and hard links so we can test better, v2

4 years agoexamples: add more deps and hard links so we can test better, v2
Michael D Labriola [Tue, 16 Jul 2013 23:44:45 +0000]
examples: add more deps and hard links so we can test better, v2

Oops, forgot to add pizza.cpp

4 years agofeatures: started plumbing for iter stages, v4
Michael D Labriola [Tue, 16 Jul 2013 21:37:20 +0000]
features: started plumbing for iter stages, v4

Added the FILES cached TarInfo list to brp.  But it doesn't work.
Apparently, you can't pickle a TarInfo object that was generated from a
TarFile object opened in write mode.  WTF?!  Testing in an interpreter with
a read-mode TarFile instance works fine...  Gonna have to rethink this a
little, because I was planning on letting the stage funcs store extra stuff
in the TarInfo objects' dicts... but they all exectute while we're
writing...

Actually, it could be the difference between the return value of
tar.getmember() vs tar.gettarinfo()...

Wait, TarInfo objects returned via gettarinfo() have a tarfile member that
needs to be removed before we can pickle it!  Adding a del(x.tarfile) call
to each item will make that list pickleable!

4 years agoMerge branch 'master' into topic/iter_stages
Michael D Labriola [Tue, 16 Jul 2013 20:30:07 +0000]
Merge branch 'master' into topic/iter_stages

* master:
  features: fixed stage_struct sorting bug
  examples: add more deps and hard links so we can test better

4 years agofeatures: fixed stage_struct sorting bug
Michael D Labriola [Tue, 16 Jul 2013 20:29:46 +0000]
features: fixed stage_struct sorting bug

We need to support at least 1 level of indirect deps (i.e., deps and
checksum may not directly relate to each other, but checksum has to happen
after core while deps has to happen before core).

4 years agoexamples: add more deps and hard links so we can test better
Michael D Labriola [Tue, 16 Jul 2013 20:27:24 +0000]
examples: add more deps and hard links so we can test better

4 years agofeatures: started plumbing for iter stages, v3
Michael D Labriola [Tue, 16 Jul 2013 20:25:26 +0000]
features: started plumbing for iter stages, v3

I've got things working in a single process for create and build stages
now.  Starting to look good.

NOTE: I made the create funcs take a work map just like all the other stage
      funcs.

NOTE: I'm trying to make create and build NOT leave a half baked file
      behind if they fail.  create works because it's done from within the
      core feature_func, but build doesn't really have a good place to
      clean up.

NOTE: I stopped using SpooledTemporaryFile.  I did some benchmarking, and
      it looks like that's actually a good bit slower than just using
      normal temporary files on disk.

NOTE: I'm now compressing the toplevel brp (as apposed to the BLOB) and I'm
      going to add a pickled TarInfo list (FILES) to the brp (but haven't
      done that yet) to speed up brp instantiation when we get to the
      install stage.

NOTE: I'm trying to plumb things to migrate easily to using multiple
      processes, but I have not implemented or tested any of the
      multiprocessing stuff yet.

4 years agoanother test script
Michael D Labriola [Fri, 12 Jul 2013 01:27:44 +0000]
another test script

doesn't look like gunzipping into a tempfile helps at all... doesn't hurt
really, so maybe it would help with 8 cores or something...

4 years agosome benchmarks and analysis
Michael D Labriola [Thu, 11 Jul 2013 21:30:56 +0000]
some benchmarks and analysis

4 years agoadded lots of test scripts...
Michael D Labriola [Thu, 11 Jul 2013 17:09:42 +0000]
added lots of test scripts...

The result of this testing is that I think the best way to speed up
installation might actually be to ditch the bzip compression.  It doesn't
look like I'm getting any benefit from extracting in parallel.  Extracting
in a single proc from a bzip compressed tarball was taking 55 seconds
whereas using gnu tar took 10 seconds.

Could the culprit be the Python bzip2 module?  Extracting via python an
uncompressed archive is just a smidge slower than using gnu tar (3-4 sec
vs 2-3).

Looks like there are indeed some bug reports written against Python's bz2
module in early 3.x releases.  Fingers crossed...

4 years agofeatures: started plumbing for iter stages, v2
Michael D Labriola [Mon, 8 Jul 2013 02:43:14 +0000]
features: started plumbing for iter stages, v2

I've added a new test.py script here, trying to experiment with the
multiprocessing module.  I wanted to use the iter stages to transparently
feed work to parallel processes... which might still work out for most of
the work I was envisioning... but I don't think it'll work for iterating
over the tmp dir and adding files to the BLOB (tarfile).  Even if I could
get the proxy stuff to work right for accessing a tarfile via shared mem,
it would be pretty stupid.  A tarfile is really supposed to be a sequece of
bytes written to tape, so we would have to serialize all the add() calls...
which would defeat the purpose.

Maybe we just don't have the install_iter stage run in parallel... or maybe
we should make each Feature be able to specify whether each of its iter
stages can be run in parallel?

4 years agofeatures: musings on iter stages, v2
Michael D Labriola [Mon, 8 Jul 2013 02:31:26 +0000]
features: musings on iter stages, v2

4 years agoMerge branch 'topic/multi' into topic/iter_stages
Michael D Labriola [Sun, 7 Jul 2013 23:57:24 +0000]
Merge branch 'topic/multi' into topic/iter_stages

* topic/multi:
  added musings regarding multiprocessing, v2
  updated docs regarding extensible features

4 years agoMerge branch 'master' into topic/multi topic/multi
Michael D Labriola [Sun, 7 Jul 2013 23:56:59 +0000]
Merge branch 'master' into topic/multi

* master:
  added musings regarding multiprocessing, v2
  updated docs regarding extensible features

4 years agoadded musings regarding multiprocessing, v2
Michael D Labriola [Sun, 7 Jul 2013 23:56:17 +0000]
added musings regarding multiprocessing, v2

4 years agoupdated docs regarding extensible features
Michael D Labriola [Sun, 7 Jul 2013 23:55:52 +0000]
updated docs regarding extensible features

4 years agofeatures: musings on iter stages
Michael D Labriola [Sun, 7 Jul 2013 23:50:38 +0000]
features: musings on iter stages

4 years agofeatures: started plumbing for iter stages
Michael D Labriola [Sun, 7 Jul 2013 17:57:53 +0000]
features: started plumbing for iter stages

4 years agofeatures/core: multiprocessing for install_func
Michael D Labriola [Sat, 6 Jul 2013 21:09:17 +0000]
features/core: multiprocessing for install_func

This is definately not done yet, but i think i want it to work something
like this.

4 years agoadded musings regarding multiprocessing
Michael D Labriola [Sat, 6 Jul 2013 14:40:53 +0000]
added musings regarding multiprocessing

4 years agofeatures/core: fleshed out install_func, v3
Michael D Labriola [Mon, 22 Apr 2013 01:22:30 +0000]
features/core: fleshed out install_func, v3

We're now installing manifest (which i'm going to go back and consistently
refer to as db) entry in /var/lib/srp/pkgname/SHA.  So far it's NOTES and
FILES.

Still a bunch of FIXMEs and half-implemented stuff...  upgrade logic, for
example is completely missing.  I should probably also go back and recreate
a fobj for the NOTES file at the very end of db entry creation so we can
update the NOTES file during installation.  And perhaps we should create
the db as yet another compressed tarball as apposed to a dir of files?

4 years agofeatures/core: fleshed out install_func, v2
Michael D Labriola [Sun, 21 Apr 2013 03:15:37 +0000]
features/core: fleshed out install_func, v2

Started playing with pickle, which looks good.  I think we'll install a
pickled list of TarInfo in /var/lib/srp/pkgname/sha.  Other features can
install other files in there as well (e.g., checksum).

4 years agocli,features/core: reworking SHA handling in package files, v2
Michael D Labriola [Sun, 21 Apr 2013 02:29:51 +0000]
cli,features/core: reworking SHA handling in package files, v2

Actually got around to testing this and ironed out the problems I knew I
would find (typos, brain farts, etc).

4 years agocli,features/core: reworking SHA handling in package files
Michael D Labriola [Thu, 18 Apr 2013 03:10:10 +0000]
cli,features/core: reworking SHA handling in package files

This is incomplete and untested... but it's late and I'm tired.

4 years agofeatures/core: fleshed out install_func
Michael D Labriola [Thu, 18 Apr 2013 03:08:55 +0000]
features/core: fleshed out install_func

So far, we're extracting the files.  We still need to record the package
manifest.

4 years agonotes: added blob_compression to check method
Michael D Labriola [Thu, 18 Apr 2013 03:07:16 +0000]
notes: added blob_compression to check method

4 years agofeatures/deps: fleshed out install_func
Michael D Labriola [Thu, 18 Apr 2013 03:06:39 +0000]
features/deps: fleshed out install_func

4 years agocli: fixed optional argument parsing
Michael D Labriola [Thu, 18 Apr 2013 03:04:48 +0000]
cli: fixed optional argument parsing

4 years agofeatures: added optional feature dependencies
Michael D Labriola [Sun, 24 Mar 2013 20:36:58 +0000]
features: added optional feature dependencies

This commit allows stage_structs to be created with features listed as deps
preceded with a question mark to indicate deps rules if a feature is
enabled, w/out recursively enabling it.  For example, if checksum is
enabled, strip_debug has to come first, but if it's not, strip_debug
doesn't care.

4 years agofeatures: added strip_debug feature module
Michael D Labriola [Sun, 24 Mar 2013 20:30:25 +0000]
features: added strip_debug feature module

4 years agonotes: added update method
Michael D Labriola [Sun, 24 Mar 2013 20:29:46 +0000]
notes: added update method

This method is used to update internal notes fields after instantiation
(e.g., cli to update features as requested via command line arguments).

4 years agofeatures: added argument to install funcs
Michael D Labriola [Sun, 24 Mar 2013 20:24:31 +0000]
features: added argument to install funcs

4 years agocli: started plumbing the install code, v2
Michael D Labriola [Sun, 24 Mar 2013 20:24:00 +0000]
cli: started plumbing the install code, v2

4 years agonotes: added check method, v2
Michael D Labriola [Mon, 4 Mar 2013 03:45:46 +0000]
notes: added check method, v2

Oops, missed the usage in do_create...

4 years agocli: started plumbing the install code
Michael D Labriola [Mon, 4 Mar 2013 03:42:02 +0000]
cli: started plumbing the install code

4 years agotoc: removed
Michael D Labriola [Mon, 4 Mar 2013 03:41:45 +0000]
toc: removed

4 years agonotes: added check method
Michael D Labriola [Mon, 4 Mar 2013 03:41:33 +0000]
notes: added check method

This new method raises an exception if srp version or feature list is
inadequate for the NOTES file being parsed.

This commit also fixes notes usage in cli.

4 years agoconfig: added config submodule
Michael D Labriola [Mon, 4 Mar 2013 03:39:02 +0000]
config: added config submodule

4 years agofeatures/deps: initial commit
Michael D Labriola [Mon, 4 Mar 2013 02:47:08 +0000]
features/deps: initial commit

This new feature has a working build func that adds deps to the NOTES file
during build.  The install_func still needs to be fleshed out

4 years agocli,features: toned down debug
Michael D Labriola [Mon, 4 Mar 2013 02:45:57 +0000]
cli,features: toned down debug

4 years agoexample: removed env from build script
Michael D Labriola [Mon, 4 Mar 2013 02:44:36 +0000]
example: removed env from build script

4 years agoexample: link against libreadline and libncurses
Michael D Labriola [Sun, 3 Mar 2013 23:51:40 +0000]
example: link against libreadline and libncurses

This commit makes our silly example executable link against a few libraries
so we can test the deps feature.

4 years agocli: working on do_build, v2
Michael D Labriola [Sun, 3 Mar 2013 12:39:11 +0000]
cli: working on do_build, v2

We're now actually writing things correctly (uncompressed tar containing
NOTES and compressed tar) and autodetecting machine architecture for the
brp name.

4 years agonotes: added write method, v2
Michael D Labriola [Sun, 3 Mar 2013 12:36:50 +0000]
notes: added write method, v2

4 years agocli: working on do_build
Michael D Labriola [Fri, 1 Mar 2013 04:06:24 +0000]
cli: working on do_build

This commit isn't finished, but does make it so we can at least write out
a misnamed tarball to verify that the perms build_func is working.

4 years agofeatures/perms: started working on build method, v3
Michael D Labriola [Fri, 1 Mar 2013 04:03:34 +0000]
features/perms: started working on build method, v3

Fixed a few minor issues.  This actually works now.

4 years agoexample: added more complex perms rules
Michael D Labriola [Fri, 1 Mar 2013 04:01:44 +0000]
example: added more complex perms rules

This commit amps up the complexity of our perms section for testing setuid,
symlinks, etc.

4 years agoexample: added symlinks to installed files
Michael D Labriola [Fri, 1 Mar 2013 03:58:51 +0000]
example: added symlinks to installed files

This commit adds a symlink to a file and a symlink to a dir to the install
target.  This is for testing behavior of srp's installation routines that
need to do special things for regular files vs dirs vs symlinks, etc.

4 years agofeatures/perms: started working on build method, v2
Michael D Labriola [Fri, 1 Mar 2013 01:15:27 +0000]
features/perms: started working on build method, v2

This appears to be all set.  Pattern matching works and payload files have
their TarInfo objects modifed as desired.

4 years agofeatures/perms: started working on build method, v1
Michael D Labriola [Thu, 28 Feb 2013 01:29:06 +0000]
features/perms: started working on build method, v1

This isn't done yet, but it's fleshed out nicely.  Moved the old 'new'
owneroverride class into the feature module (named it perms) and set up
the build_func to iterate over payload files examining the regular
expressions from the NOTES file's perms section.  Matching seems to work,
now to actually modify the manifest's TarInfo objects.