Add an XSL stylesheet for listing packages in LFS. Those are
not listed anywhere else, but they are considered dependencies of
everything, so should never be erased.
For outputting DESTDIR=, we need to test whether the text contains
"make". But when it contains "qmake" (or any different word containing
"make" but not exactly "make), we don't want to output anything more
than the text. So strengthen the test.
We output text()[1], literal (literally), text(2). This does
not work if there are several <literal> tags. Change it to
outputing preceding-sibling::text()[1], literal for each
literal tag, then output text()[last()] (which works also
when there are no literal tag, so it removes a choose.
For additional downloads, the normal book layout is a <bridgehead>
containing "Additional Downloads" followed by one or more
<itemizedlist>, one for each file to download. But on some pages,
there are several <bridgehead>, each followed by one or more
<itemizedlist>. So we have to check that only the <itemizedlist>
correponding to the current <bridgehead> are treated. Otherwise,
the download code may be generated more than once. This imposes
that no other bridgehead is inserted between the one containing
"Additional" (or in some case "Recommended/Optional Download") and
the <itemizedlist>.
Instead of having them inside the Makefile, which implies editing
them (automatically), which is not robust if anything changes:
it got broken when a test for a mountpoint was added...
Presently, when the <node>groupxx file is created, we copy "a"
lines from <node> to <node>groupxx (changing "a" to "b") and
we remove all the lines containing the runtime dependency from
<node>. This may not always be what we want: for example, a
dependency can be both "required at runtime" and "optional for
tests". We should keep this information when cleaning the nodes.
Only when removing circular paths should the optional tie be broken
if necessary.
When not downloading packages, the function "get_sources" returns
early because it begins with:
[ "$GETPKG" = y ] || return
The problem is that in this case it returns with an exit code
inherited from the last executed instruction, which is the
test (and therefore "false"). The solution is to change to
return 0.
- update versions and instructions in packageManager.xml.pacman.
Also remove the pages for old books.
- Fix packInstall.sh.pacman, so that restarting after a failure
is possible
- Change LFS/lfs.xsl so that destdir install use symlinks for
/lib, /sbin, and /bin. Remove also absolete commands for old
books
- fix gcc script in book parser
LFS/lfs.xsl does not generate correct commands for stripping if
STRIP=n. LFS/master.sh had already been corrected to not run
stripping if STRIP=n, but only for the first pass. Do this for
all passes.
In order to be able to pass NINJAJOBS=1 when optimize is 0, we
add two paramters to this function. The three parameters are
then:
- the name of the package (tested against opt_override)
- MAKEFLAGS for this package
- NINJAJOBS for this package
Also fix typos and shell programming. Use "cat file | while read"
instead of a complicated construct for reading only lines and then
breaking lines into their fields... Fix a TODO item
Now that we have only lfs, some lines in the Makefile have been
removed, but still some shell continuation (&& \) were left. This
leads to weird error message (@ non existing command, or so).
Remove the continuation characters.
- do not try to copy downloaded files to $SRC_ARCHIVE: they may
belong to another user, and so be non writable.
- trunk/BOOK is not used anymore in BLFS/Makefile
- do not make update a second time
- If MISSING_FILES.DMP is left from a preceding attempt, it may
generate a false warning a bout missing files even if they could
all be downloaded.
- do not try to copy a modified file to SRC_ARCHIVE if it already
exists: it may belong to another user...
