Open zarniwhoop73 opened 8 years ago
Thanks for bringing this to my attention.
I contemplated about your awk-usage proposal, and decided against it. We cannot assume that awk is available on Windows or MacOS. I think I'll use perl for that action in Makefile; perl is already a hard dependency both in build and use of xindy, and this should not decrease portability.
I gather that you are xindy's downstream for LFS. Where do you get xindy from? From here (github)? or from TeX-Live's SVN? I ask, because sometimes 3rd party portability patches arrive in TL first and I'd like to know how important it is to add them prompto to this repository.
Again, thank's for engaging me. Joachim
On Mon, Feb 29, 2016 at 05:52:46PM -0800, Joachim Schrod wrote:
Thanks for bringing this to my attention.
I contemplated about your awk-usage proposal, and decided against it. We cannot assume that awk is available on Windows or MacOS. I think I'll use perl for that action in Makefile; perl is already a hard dependency both in build and use of xindy, and this should not decrease portability.
I gather that you are xindy's downstream for LFS. Where do you get xindy from? From here (github)? or from TeX-Live's SVN? I ask, because sometimes 3rd party portability patches arrive in TL first and I'd like to know how important it is to add them prompto to this repository.
Again, thank's for engaging me. Joachim
I saw all the perl and assumed you would probably come up with something in perl. We (BLFS) tend to only use releases, and to only patch them when we need to, so for the moment we are taking 2.5.1 from tug.ctan.org//support/xindy/base - for TeX we only really upgrade once a year, but the "extras" not included in basic texlive-source get upgraded when we notice a new release.
Now that it is building again, we probably won't touch it until the next release.
Regards,
This email was written using 100% recycled letters.
Grep-2.23 treats all files not in the current encoding as binary. The latin1, latin2, latin3 files are therefore victims of this (instead of the expected content, the file just contains an extra blank line after the header, followed by ' Binary file (standard input) matches' because each has a different encoding.
In Beyond Linuxfromscratch we already force the C|POSIX locale to placate the sort program in recent coreutils. Therefore when the build tries to create utf8.xdy that is similarly damaged.
The use of '"grep -v '^;'" is only to remove blank lines. I took what seemed to be a simple approach (using grep -a might work, but is possibly dependant on the flavour of grep installed - I'm thinking particularly of minimalist variants), so I just changed that to "awk NF" in make-rules/inputenc/Makefile.in : obviously whatever you decide to do needs to happen in the .ac version.