| Age | Commit message (Collapse) | Author |
|---|
|
|
|
This is just adding another cond-expand
|
|
|
|
When moving files from the temporary directory to the cache directory,
we must handle symlinks first. If we handle the _target_ before the
symlink, the target will be deleted and `copy-file' will break when
trying to copy the symlink, as it will be dangling.
Signed-off-by: Peter Bex <peter@more-magic.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Forgot about the fact that hooks exist.
It is unclear how this should work with initfiles that register hooks.
Ideally it would simply be a matter of exporting the hooks, but then
we need a C4 installation procedure that somehow installs just the
program and an import library. With C5 this would be easier, but I
don't yet want to make henrietta-cache C5-only.
|
|
|
|
|
|
|
|
This makes it easier to find out missing imports.
|
|
This used lset-intersection but that should've been lset-difference.
|
|
The -k or -keep option tells henrietta-cache to throw away older
releases, so that only the supplied number of latest releases is kept.
The -n option tells henrietta-cache to not download any new releases.
It can be used if one wants to prune old releases, together with -k.
|
|
Somehow this now causes an exception to be raised when the temp dir
was the contents dir, because move-to-cache already removes that
directory.
|
|
|
|
|
|
tmp causes warnings when linking on OpenBSD. Group ownership switches
to the directory and warns when moving (on OpenBSD and possibly other
BSDs, too).
Thanks to "n0goOi3"
|
|
|
|
The -i (init file) option means we can't wrap things in a module,
unless we build and ship the module, which will mean people need
to rewrite the init file. Not worth the hassle to figure out how
to do it "properly".
Fixes #1531, reported by Mario Goulart
|
|
|
|
This drops support for CHICKEN versions before 4.6 (copy-port is required now)
|
|
Fix some issues like depending on srfi-13 implicitly only for
string-null? and some other bad imports.
|
|
egg-files and "files" in meta-files
There's some minor code duplication but it's worth the loss of
confusion and potential for error.
|
|
Support both `meta-file' and `egg-file' as URI types, and both `files'
and `distribution-files' forms in egg description files (.meta for C4
and .egg for C5).
Note that:
* Although `distribution-files' is intended to be used by C5's
chicken-install, it can also be used in the toplevel of .meta files,
as chicken-install for C4 is very permissive. `files', however,
cannot be used in the toplevel of .egg files.
* (uri meta-file ...) can point to a .egg file, and (uri egg-file ...)
can point to a .meta file. `egg-file' is just and alias to
`meta-file' for naming consistency in C5.
* henrietta-cache for C4 can be used to cache eggs for C5 (and
supports `egg-file' and `distribution-files').
Signed-off-by: Peter Bex <peter@more-magic.net>
|
|
|
|
|
|
|
|
|
|
As reported by Jim Ursetto, the version of zcat shipped with OS X doesn't handle gzipped files, and gzcat doesn't exist on many Linuxen. However, gunzip has a -c option which does the same. For consistency, use bunzip2 -c as well.
|
|
|
|
|
|
|
|
|
|
think). Remove "tags" subdirectories
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This ensures that when we crap out mid-way for some reason, we won't
leave an empty or half-populated directory around, which would cause
it to skip the dir instead of trying again on the next run.
(File-list types are meta-file or plaintext flat file listing)
|
|
Pax supports replacement patterns which makes it slightly simpler to
use (and it's actually a standard), we'll just rely on the commonly
supported tar flags "x" and "f", and just hope it'll work everywhere.
This is because (at least on NetBSD), pax only supports plain ustar
format, which has silly limitations on file name length. Longer file
names are a GNU extension, which means we'll need something that
supports these extensions anyway. On the long run we should probably
switch to libarchive, but I don't feel like writing yet another egg
just to work around these few problems.
Also, don't create the target directory in advance but let the
procedures do this, so if something breaks we won't get empty
directories. This ensures it will be retried later.
|
|
|
|
|
|
|
|
|
