commit 4e72075fba introduced late locking for top level packages,
but used the wrong package's top level flag for the check. this
fixes a problem that dependencies might not get pulled in.
move all files therein to other places. this allows /var to be
mounted from harddisk, but rest of system be run from ramdisk.
this also removes support for historical version of the scripts
database which was obsoleted in 2.0_pre16 (in July 2009).
If GNU wget is available, then we want use it as it gives the user
possibility to set misc options, such as bandwidth limit in wgetrc.
It might also be that busybox is not available in case bootstrapping
alpine from other distros (think debian vserver host creating an
alpine guest)
In any case we, fall back to busybox wget in case wget libs are getting
upgraded. (see http://redmine.alpinelinux.org/issues/347)
Packages without architecture should get the platforms default
architecture (so we are backwards compatible). Only the virtual
packages should get 'noarch' by default. Also print full path
to the index file which failed.
Parse install_if from package metadata and include it in the
indexes. Also setup the reverse install_if dependencies when
loading a database. ref #443.
Actual install_if functionality is not yet implemented.
Just disable installation of packages using the new stuff. Also
flag lower case package info fields as non-critical and allow
installation even if that features is not supported.
When package is installed from commandline, we should always
install that specific instance of package (never favor repository
version if it has difference identity). Otherwise we might not
always end-up installing the .apk given on command line. The
dependency is now against specific checksum identity (marked
with >< dependency comparison). Fixes#492.
I think there used to be problems with using http_proxy on bb wget,
but those have been fixed for quite some time. This fixes#347.
We should probably use libcurl or similar library eventually, but
I'm not entirely sure if/when we want that.
The boundary callback should not happen until all the uncompressed
data has been consumed. This previously seems to have worked
because normally gzip library returns "no error" instead of the
"stream end" if we extract exactly the amount of bytes remaining
in the archive. (Perhaps this was changed in new zlib.) In any
case, verification was broken with some apks due to this callback
ordering issue.
- implement a hash table for commonly shared fields such as
license, version and architecture
- use macroes to print blobs or pkgname-pkgver strings
- fix some old cruft
The hash is 'unsigned long' everywhere which is 64-bits (on 64-bit
boxes). Fix the one instance of it being 'unsigned int' since that
would be 32-bit and cause lookup failures on 64-bit boxes.
We want remove dirs when they are unreferenced so we remove all dirs on
apk del, but we don't want remove dirs when closing database. So we make
removing dir optional when unreferencing it.
This partially reverts commit c7ffc96a16.
fixes#406
Timo Teräs