2 commits - createrepo/__init__.py createrepo/utils.py genpkgmetadata.py mergerepo.py modifyrepo.py

[s***@osuosl.org]

createrepo/__init__.py | 83 +++++++++++++++++++++++++------------------------
createrepo/utils.py | 43 +++++++++++++++++++++++++
genpkgmetadata.py | 8 ++++
mergerepo.py | 8 ++--
modifyrepo.py | 29 +++++++++++++----
5 files changed, 122 insertions(+), 49 deletions(-)

New commits:
commit 6715fb5a87d2255660daa42748081582ae59a330
Merge: dafea8c 8c35f58
Author: Seth Vidal <***@fedoraproject.org>
Date: Fri Sep 16 01:37:33 2011 -0400

Merge branch 'master' of ssh://createrepo.baseurl.org/srv/projects/createrepo/git/createrepo

* 'master' of ssh://createrepo.baseurl.org/srv/projects/createrepo/git/createrepo:
Add --xz to createrepo and mergerepo bash completions.

commit dafea8c15d5feaaf50a9df4898fd34eed1b09916
Author: Seth Vidal <***@fedoraproject.org>
Date: Thu Sep 15 17:55:50 2011 -0400

add --compress-type everywhere and make it all behave.

can't default to xz yet for all md b/c of y-m-p silliness.

diff --git a/createrepo/__init__.py b/createrepo/__init__.py
index 7d147a9..61694f9 100644
--- a/createrepo/__init__.py
+++ b/createrepo/__init__.py
@@ -34,7 +34,7 @@ from yum.packageSack import MetaSack
from yum.packages import YumAvailablePackage

import rpmUtils.transaction
-from utils import _, errorprint, MDError, lzma
+from utils import _, errorprint, MDError, lzma, _available_compression
import readMetadata
try:
import sqlite3 as sqlite
@@ -46,7 +46,7 @@ try:
except ImportError:
pass

-from utils import _gzipOpen, bzipFile, xzFile, checkAndMakeDir, GzipFile, \
+from utils import _gzipOpen, compressFile, compressOpen, checkAndMakeDir, GzipFile, \
checksum_and_rename, split_list_into_equal_chunks
import deltarpms

@@ -74,7 +74,7 @@ class MetaDataConfig(object):
self.deltadir = None
self.delta_relative = 'drpms/'
self.oldpackage_paths = [] # where to look for the old packages -
- self.deltafile = 'prestodelta.xml.gz'
+ self.deltafile = 'prestodelta.xml'
self.num_deltas = 1 # number of older versions to delta (max)
self.max_delta_rpm_size = 100000000
self.update_md_path = None
@@ -86,9 +86,9 @@ class MetaDataConfig(object):
self.skip_symlinks = False
self.pkglist = []
self.database_only = False
- self.primaryfile = 'primary.xml.gz'
- self.filelistsfile = 'filelists.xml.gz'
- self.otherfile = 'other.xml.gz'
+ self.primaryfile = 'primary.xml'
+ self.filelistsfile = 'filelists.xml'
+ self.otherfile = 'other.xml'
self.repomdfile = 'repomd.xml'
self.tempdir = '.repodata'
self.finaldir = 'repodata'
@@ -110,7 +110,8 @@ class MetaDataConfig(object):
self.worker_cmd = '/usr/share/createrepo/worker.py'
#self.worker_cmd = './worker.py' # helpful when testing
self.retain_old_md = 0
- self.xz = False # use xz for compression
+ self.compress_type = 'gz'
+

class SimpleMDCallBack(object):
def errorlog(self, thing):
@@ -146,8 +147,13 @@ class MetaDataGenerator:
if not self.conf.directory and not self.conf.directories:
raise MDError, "No directory given on which to run."

- if self.conf.xz and not utils.lzma:
- raise MDError, "XZ compression requested but lzma/xz module not available."
+ if not self.conf.compress_type:
+ self.conf.compress_type = 'gz'
+
+ if self.conf.compress_type not in utils._available_compression:
+ raise MDError, "Compression %s not available: Please choose from: %s" \
+ % (self.conf.compress_type, ', '.join(utils._available_compression))
+

if not self.conf.directories: # just makes things easier later
self.conf.directories = [self.conf.directory]
@@ -414,9 +420,11 @@ class MetaDataGenerator:

def _setupPrimary(self):
# setup the primary metadata file
+ # FIXME - make this be conf.compress_type once y-m-p is fixed
+ fpz = self.conf.primaryfile + '.' + 'gz'
primaryfilepath = os.path.join(self.conf.outputdir, self.conf.tempdir,
- self.conf.primaryfile)
- fo = _gzipOpen(primaryfilepath, 'w')
+ fpz)
+ fo = compressOpen(primaryfilepath, 'w', 'gz')
fo.write('<?xml version="1.0" encoding="UTF-8"?>\n')
fo.write('<metadata xmlns="http://linux.duke.edu/metadata/common"' \
' xmlns:rpm="http://linux.duke.edu/metadata/rpm" packages="%s">' %
@@ -425,9 +433,11 @@ class MetaDataGenerator:

def _setupFilelists(self):
# setup the filelist file
+ # FIXME - make this be conf.compress_type once y-m-p is fixed
+ fpz = self.conf.filelistsfile + '.' + 'gz'
filelistpath = os.path.join(self.conf.outputdir, self.conf.tempdir,
- self.conf.filelistsfile)
- fo = _gzipOpen(filelistpath, 'w')
+ fpz)
+ fo = compressOpen(filelistpath, 'w', 'gz')
fo.write('<?xml version="1.0" encoding="UTF-8"?>\n')
fo.write('<filelists xmlns="http://linux.duke.edu/metadata/filelists"' \
' packages="%s">' % self.pkgcount)
@@ -435,9 +445,11 @@ class MetaDataGenerator:

def _setupOther(self):
# setup the other file
+ # FIXME - make this be conf.compress_type once y-m-p is fixed
+ fpz = self.conf.otherfile + '.' + 'gz'
otherfilepath = os.path.join(self.conf.outputdir, self.conf.tempdir,
- self.conf.otherfile)
- fo = _gzipOpen(otherfilepath, 'w')
+ fpz)
+ fo = compressOpen(otherfilepath, 'w', 'gz')
fo.write('<?xml version="1.0" encoding="UTF-8"?>\n')
fo.write('<otherdata xmlns="http://linux.duke.edu/metadata/other"' \
' packages="%s">' %
@@ -446,9 +458,10 @@ class MetaDataGenerator:

def _setupDelta(self):
# setup the other file
+ fpz = self.conf.deltafile + '.' + self.conf.compress_type
deltafilepath = os.path.join(self.conf.outputdir, self.conf.tempdir,
- self.conf.deltafile)
- fo = _gzipOpen(deltafilepath, 'w')
+ fpz)
+ fo = compressOpen(deltafilepath, 'w', self.conf.compress_type)
fo.write('<?xml version="1.0" encoding="UTF-8"?>\n')
fo.write('<prestodelta>\n')
return fo
@@ -612,7 +625,6 @@ class MetaDataGenerator:

for worker_num in range(self.conf.workers):
pkl = self._worker_tmp_path + '/pkglist-%s' % worker_num
- print pkl
f = open(pkl, 'w')
f.write('\n'.join(worker_chunks[worker_num]))
f.close()
@@ -828,7 +840,7 @@ class MetaDataGenerator:
return ' '.join(results)

def _createRepoDataObject(self, mdfile, mdtype, compress=True,
- compress_type='gzip', attribs={}):
+ compress_type=None, attribs={}):
"""return random metadata as RepoData object to be added to RepoMD
mdfile = complete path to file
mdtype = the metadata type to use
@@ -838,19 +850,12 @@ class MetaDataGenerator:
sfile = os.path.basename(mdfile)
fo = open(mdfile, 'r')
outdir = os.path.join(self.conf.outputdir, self.conf.tempdir)
+ if not compress_type:
+ compress_type = self.conf.compress_type
if compress:
- if compress_type == 'gzip':
- sfile = '%s.gz' % sfile
- outfn = os.path.join(outdir, sfile)
- output = GzipFile(filename = outfn, mode='wb')
- elif compress_type == 'bzip2':
- sfile = '%s.bz2' % sfile
- outfn = os.path.join(outdir, sfile)
- output = BZ2File(filename = outfn, mode='wb')
- elif compress_type == 'xz':
- sfile = '%s.xz' % sfile
- outfn = os.path.join(outdir, sfile)
- output = utils.lzma.LZMAFile(outfn, mode='wb')
+ sfile = '%s.%s' % (sfile, compress_type)
+ outfn = os.path.join(outdir, sfile)
+ output = compressOpen(outfn, mode='wb', compress_type=compress_type)

else:
outfn = os.path.join(outdir, sfile)
@@ -924,9 +929,13 @@ class MetaDataGenerator:
rp = sqlitecachec.RepodataParserSqlite(repopath, repomd.repoid, None)

for (rpm_file, ftype) in workfiles:
+ # when we fix y-m-p and non-gzipped xml files - then we can make this just add
+ # self.conf.compress_type
+ if ftype in ('other', 'filelists', 'primary'):
+ rpm_file = rpm_file + '.' + 'gz'
complete_path = os.path.join(repopath, rpm_file)

- zfo = _gzipOpen(complete_path)
+ zfo = compressOpen(complete_path)
# This is misc.checksum() done locally so we can get the size too.
data = misc.Checksums([sumtype])
while data.read(zfo, 2**16):
@@ -967,17 +976,13 @@ class MetaDataGenerator:

# rename from silly name to not silly name
os.rename(tmp_result_path, resultpath)
- ext = 'bz2'
- compress_func = bzipFile
- if self.conf.xz:
- ext = 'xz'
- compress_func = xzFile
+ ext = self.conf.compress_type
compressed_name = '%s.%s' % (good_name, ext)
result_compressed = os.path.join(repopath, compressed_name)
db_csums[ftype] = misc.checksum(sumtype, resultpath)

# compress the files
- compress_func(resultpath, result_compressed)
+ compressFile(resultpath, result_compressed, self.conf.compress_type)
# csum the compressed file
db_compressed_sums[ftype] = misc.checksum(sumtype,
result_compressed)
@@ -1051,7 +1056,7 @@ class MetaDataGenerator:

if self.conf.additional_metadata:
for md_type, md_file in self.conf.additional_metadata.items():
- mdcontent = self._createRepoDataObject(md_file, md_type, compress_type='xz')
+ mdcontent = self._createRepoDataObject(md_file, md_type)
repomd.repoData[mdcontent.type] = mdcontent


diff --git a/createrepo/utils.py b/createrepo/utils.py
index 655083d..c816640 100644
--- a/createrepo/utils.py
+++ b/createrepo/utils.py
@@ -91,6 +91,49 @@ def xzFile(source, dest):
destination.close()
s_fn.close()

+def gzFile(source, dest):
+
+ s_fn = open(source, 'rb')
+ destination = GzipFile(dest, 'w')
+
+ while True:
+ data = s_fn.read(1024000)
+
+ if not data: break
+ destination.write(data)
+
+ destination.close()
+ s_fn.close()
+
+
+
+def compressFile(source, dest, compress_type):
+ """Compress an existing file using any compression type from source to dest"""
+
+ if compress_type == 'xz':
+ xzFile(source, dest)
+ elif compress_type == 'bz2':
+ bzipFile(source, dest)
+ elif compress_type == 'gz':
+ gzFile(source, dest)
+ else:
+ raise MDError, "Unknown compression type %s" % compress_type
+
+def compressOpen(fn, mode='rb', compress_type=None):
+
+ if not compress_type:
+ # we are readonly and we don't give a compress_type - then guess based on the file extension
+ compress_type = fn.split('.')[-1]
+
+ if compress_type == 'xz':
+ return lzma.LZMAFile(fn, mode)
+ elif compress_type == 'bz2':
+ return bz2.BZ2File(fn, mode)
+ elif compress_type == 'gz':
+ return _gzipOpen(fn, mode)
+ else:
+ raise MDError, "Unknown compression type %s" % compress_type
+
def returnFD(filename):
try:
fdno = os.open(filename, os.O_RDONLY)
diff --git a/genpkgmetadata.py b/genpkgmetadata.py
index 4ba445e..af0ecb4 100755
--- a/genpkgmetadata.py
+++ b/genpkgmetadata.py
@@ -127,6 +127,9 @@ def parse_args(args, conf):
parser.add_option("--xz", default=False,
action="store_true",
help="use xz for repodata compression")
+ parser.add_option("--compress-type", default=None, dest="compress_type",
+ help="which compression type to use")
+

(opts, argsleft) = parser.parse_args(args)
if len(argsleft) > 1 and not opts.split:
@@ -159,6 +162,11 @@ def parse_args(args, conf):

if opts.nodatabase:
opts.database = False
+
+ # xz is just a shorthand for compress_type
+ if opts.xz and not opts.compress_type:
+ opts.compress_type='xz'
+

# let's switch over to using the conf object - put all the opts into it
for opt in parser.option_list:
diff --git a/mergerepo.py b/mergerepo.py
index 882395a..80cb1a8 100755
--- a/mergerepo.py
+++ b/mergerepo.py
@@ -48,8 +48,8 @@ def parse_args(args):
help="Do not merge group(comps) metadata")
parser.add_option("", "--noupdateinfo", default=False, action="store_true",
help="Do not merge updateinfo metadata")
- parser.add_option("", "--xz", default=False, action="store_true",
- help="Use xz for repodata compression")
+ parser.add_option("--compress-type", default=None, dest="compress_type",
+ help="which compression type to use")

(opts, argsleft) = parser.parse_args(args)

@@ -81,8 +81,8 @@ def main(args):
rmbase.groups = False
if opts.noupdateinfo:
rmbase.updateinfo = False
- if opts.xz:
- rmbase.mdconf.xz = True
+ if opts.compress_type:
+ rmbase.mdconf.compress_type = opts.compress_type
try:
rmbase.merge_repos()
rmbase.write_metadata()
diff --git a/modifyrepo.py b/modifyrepo.py
index c3370e8..153ad4d 100755
--- a/modifyrepo.py
+++ b/modifyrepo.py
@@ -29,7 +29,7 @@
import os
import sys
from createrepo import __version__
-from createrepo.utils import checksum_and_rename, GzipFile, MDError
+from createrepo.utils import checksum_and_rename, compressOpen, MDError
from yum.misc import checksum

from yum.repoMDObject import RepoMD, RepoMDError, RepoData
@@ -44,6 +44,8 @@ class RepoMetadata:
self.repodir = os.path.abspath(repo)
self.repomdxml = os.path.join(self.repodir, 'repomd.xml')
self.checksum_type = 'sha256'
+ self.compress = False
+ self.compress_type='xz'

if not os.path.exists(self.repomdxml):
raise MDError, '%s not found' % self.repomdxml
@@ -97,8 +99,8 @@ class RepoMetadata:
mdname = 'updateinfo.xml'
elif isinstance(metadata, str):
if os.path.exists(metadata):
- if metadata.endswith('.gz'):
- oldmd = GzipFile(filename=metadata, mode='rb')
+ if metadata.split('.')[-1] in ('gz', 'bz2', 'xz'):
+ oldmd = compressOpen(metadata, mode='rb')
else:
oldmd = file(metadata, 'r')
md = oldmd.read()
@@ -109,13 +111,19 @@ class RepoMetadata:
else:
raise MDError, 'invalid metadata type'

+ do_compress = False
## Compress the metadata and move it into the repodata
- if not mdname.endswith('.gz'):
- mdname += '.gz'
+ if self.compress or not mdname.split('.')[-1] in ('gz', 'bz2', 'xz'):
+ do_compress = True
+ mdname += '.' + self.compress_type
mdtype = self._get_mdtype(mdname, mdtype)

destmd = os.path.join(self.repodir, mdname)
- newmd = GzipFile(filename=destmd, mode='wb')
+ if do_compress:
+ newmd = compressOpen(destmd, mode='wb', compress_type=self.compress_type)
+ else:
+ newmd = open(destmd, 'wb')
+
newmd.write(md)
newmd.close()
print "Wrote:", destmd
@@ -166,6 +174,10 @@ def main(args):
help="specific datatype of the metadata, will be derived from the filename if not specified")
parser.add_option("--remove", action="store_true",
help="remove specified file from repodata")
+ parser.add_option("--compress", action="store_true", default=False,
+ help="compress the new repodata before adding it to the repo")
+ parser.add_option("--compress-type", dest='compress_type', default='xz',
+ help="compression format to use")
parser.usage = "modifyrepo [options] [--remove] <input_metadata> <output repodata>"

(opts, argsleft) = parser.parse_args(args)
@@ -180,6 +192,10 @@ def main(args):
print "Could not access repository: %s" % str(e)
return 1

+
+ repomd.compress = opts.compress
+ repomd.compress_type = opts.compress_type
+
# remove
if opts.remove:
try:
@@ -195,6 +211,7 @@ def main(args):
except MDError, e:
print "Could not add metadata from file %s: %s" % (metadata, str(e))
return 1
+

if __name__ == '__main__':
ret = main(sys.argv[1:])

[Anders F Björklund]

Post by s***@osuosl.org
commit dafea8c15d5feaaf50a9df4898fd34eed1b09916
Date: Thu Sep 15 17:55:50 2011 -0400
add --compress-type everywhere and make it all behave.
can't default to xz yet for all md b/c of y-m-p silliness.

There was some patches* for yum-metadata-parser last year,
to handle the lzma and xz compression in addition to gzip ?
Since those aren't built-in to libxml2, it needed to use
a temporary file and uncompress before calling parsefile...

If you're accepting patches now, could try and dig it up.

* In the end, it ended up not being used since the new
files used primary_lzma etc and the gzip was kept too.
When the xz compression *was* used, the sqlite wasn't,
so it I don't think sqlitecachec ended up being patched.

But it should be relatively simple to add lzma/xz to it.

--anders

[seth vidal]

Post by Anders F Björklund

Post by s***@osuosl.org
commit dafea8c15d5feaaf50a9df4898fd34eed1b09916
Date: Thu Sep 15 17:55:50 2011 -0400
add --compress-type everywhere and make it all behave.
can't default to xz yet for all md b/c of y-m-p silliness.

There was some patches* for yum-metadata-parser last year,
to handle the lzma and xz compression in addition to gzip ?
Since those aren't built-in to libxml2, it needed to use
a temporary file and uncompress before calling parsefile...
If you're accepting patches now, could try and dig it up.

I don't know of when I wasn't accepting patches

Post by Anders F Björklund
* In the end, it ended up not being used since the new
files used primary_lzma etc and the gzip was kept too.
When the xz compression *was* used, the sqlite wasn't,
so it I don't think sqlitecachec ended up being patched.

Right - I'm not interested in adding repodata to handle the formats.


-sv

[Anders F Björklund]

Post by seth vidal

Post by Anders F Björklund
There was some patches* for yum-metadata-parser last year,
to handle the lzma and xz compression in addition to gzip ?
Since those aren't built-in to libxml2, it needed to use
a temporary file and uncompress before calling parsefile...
If you're accepting patches now, could try and dig it up.

I don't know of when I wasn't accepting patches

Maybe the metadata files wasn't "ready" for it, last year.
But it was with the --lzma and --xz options to createrepo.

There was another (unrelated) patch with <rpm:disttag> and
<rpm:distepoch>, so that it would work with Mandriva 2011.

Maybe they got intermixed ?

Anyway, seems it was just a hack (spawning an "unxz"...)

Probably needs something better, though I don't know if
libxml2 has a stream-like interface or just file/memory ?

Post by seth vidal

Post by Anders F Björklund
* In the end, it ended up not being used since the new
files used primary_lzma etc and the gzip was kept too.
When the xz compression *was* used, the sqlite wasn't,
so it I don't think sqlitecachec ended up being patched.

Right - I'm not interested in adding repodata to handle the formats.

As long as it doesn't need to be compatible, using the
file extension should be good enough. And xz has magic.

I added the other stuff to the stable series, instead.
But the yum patches should still be around (if needed).

--anders

[seth vidal]

Post by Anders F Björklund
I added the other stuff to the stable series, instead.
But the yum patches should still be around (if needed).

That 'stable' series you added things to is not really in use anymore.
It's labeled as the maintenance branch since it is in maintenance mode
only. No new features will be added there.

If you're going to fork it like that I'd appreciate it if you made a
different name for it. It's confusing to users.


thanks,
-sv

[Anders F Björklund]

Post by seth vidal

Post by Anders F Björklund
I added the other stuff to the stable series, instead.
But the yum patches should still be around (if needed).

That 'stable' series you added things to is not really in use anymore.
It's labeled as the maintenance branch since it is in maintenance mode
only. No new features will be added there.
If you're going to fork it like that I'd appreciate it if you made a
different name for it. It's confusing to users.

It was more of backports, but whatever. Adding changelog_limit
and that collapse_glibc_requires flag that was fooling smart,
in addition to the sha256 checksums and the lzma compression.
Plus the compatibility patches required to make it run at all.

There's nothing incompatible about the generated "repodata",
except for those extra tags (primary_lzma and rpm:distepoch) ?

Still seemed better than doing a separate rpm metadata format...

--anders

[seth vidal]

Post by Anders F Björklund

Post by seth vidal

Post by Anders F Björklund
I added the other stuff to the stable series, instead.
But the yum patches should still be around (if needed).

That 'stable' series you added things to is not really in use anymore.
It's labeled as the maintenance branch since it is in maintenance mode
only. No new features will be added there.
If you're going to fork it like that I'd appreciate it if you made a
different name for it. It's confusing to users.

It was more of backports, but whatever. Adding changelog_limit
and that collapse_glibc_requires flag that was fooling smart,
in addition to the sha256 checksums and the lzma compression.
Plus the compatibility patches required to make it run at all.
There's nothing incompatible about the generated "repodata",
except for those extra tags (primary_lzma and rpm:distepoch) ?
Still seemed better than doing a separate rpm metadata format...

I wasn't asking you to change the format.

Just change the program name.

I think calling it createrepo is just going to cause confusion.

-sv

[Anders F Björklund]

Post by seth vidal

Post by Anders F Björklund

Post by seth vidal
That 'stable' series you added things to is not really in use anymore.
It's labeled as the maintenance branch since it is in maintenance mode
only. No new features will be added there.
If you're going to fork it like that I'd appreciate it if you made a
different name for it. It's confusing to users.

It was more of backports, but whatever. Adding changelog_limit
and that collapse_glibc_requires flag that was fooling smart,
in addition to the sha256 checksums and the lzma compression.
Plus the compatibility patches required to make it run at all.
There's nothing incompatible about the generated "repodata",
except for those extra tags (primary_lzma and rpm:distepoch) ?
Still seemed better than doing a separate rpm metadata format...

I wasn't asking you to change the format.
Just change the program name.
I think calling it createrepo is just going to cause confusion.

Is that really needed ?

Like: the openSUSE program has the same name, even if it adds
a few new tags for the weak dependencies (enhances/suggests) -
and the rest are just backports of features already existing.

The only remaining patch is the fixed epoch/distepoch parsing ?
Without it, the distepoch (:2011) is included in the release...
Hardly seems worth to rename the program, for such a bugfix.

What is the confusion ?

--anders

[seth vidal]

Post by Anders F Björklund
Is that really needed ?
Like: the openSUSE program has the same name, even if it adds
a few new tags for the weak dependencies (enhances/suggests) -
and the rest are just backports of features already existing.
The only remaining patch is the fixed epoch/distepoch parsing ?
Without it, the distepoch (:2011) is included in the release...
Hardly seems worth to rename the program, for such a bugfix.
What is the confusion ?

You're creating a different set of repodata in primary/repomd.xml if
nothing else.

Now - someone using suse/mandriva will find a problem with how it makes
those and, when googling for where to report it, file a bug at
createrepo.baseurl.org.

that's what I mean by confusion. It's the same problem I have whenever
there are patches into red hat or fedora yum/createrepo that don't make
it into upstream. This is why we try to never have those.

Do you see what I mean?

-sv

[Anders F Björklund]

Post by seth vidal

Post by Anders F Björklund
What is the confusion ?

You're creating a different set of repodata in primary/repomd.xml if
nothing else.

Not sure I follow. It will either use gz or xz just like the new one,
the only difference is when you want compatibility mode and "both" ?
But you would have some problems anyway, if trying to read the lzma
compressed metadata as uncompressed (for instance). So seemed better ?

Post by seth vidal
Now - someone using suse/mandriva will find a problem with how it makes
those and, when googling for where to report it, file a bug at
createrepo.baseurl.org.

BTW: that tracker looks spammed ?

http://createrepo.baseurl.org/report/1

Post by seth vidal
that's what I mean by confusion. It's the same problem I have whenever
there are patches into red hat or fedora yum/createrepo that don't make
it into upstream. This is why we try to never have those.

I'm sure that you have compatibility and bugfix patches in fedora,
without forking and renaming ? If you're talking about backporting
features, that is slightly different... It was only done because it
was simpler than trying to port all of yum and keep it updated there.

The official standpoint is still preferring the "master" createrepo,
which in practice means that it is broken at runtime because of yum.
Using the "maintenance" createrepo was more about getting it running.
Eventually I'm sure that yum could be updated, so that it works again.

As far as I know, it's working in openSUSE even if kept at yum 3.2.25
and I think that yum 3.2.29 should be more or less fixed for Mandriva
and for FreeBSD (once deployed) which should allow createrepo 0.9.9.
I don't think anyone tried Solaris recently, but it had some issues...

Post by seth vidal
Do you see what I mean?

Yes, but not how it applies here.

I know that the --lzma was rejected upstream and replaced by --xz,
and I know that yum doesn't want to support all versions of rpm / os.
But one could still add xz compression to createrepo and sqlitecachec,
preferably upstream too, without having to call it something else ?

I don't really care, the patches are there if anyone wants them...

http://afb.users.sourceforge.net/repodata/patches/ ("maintenance")

--anders