Marius Bakke ©ó 2017-05-01 01:10 ¼g¹D:
> Hi Brendan, thanks for this!
>
> Brendan Tildesley <brendan.tildesley@openmailbox.org> writes:
>
>> * gnu/packages/fonts.scm (font-cns11643): New variable.
> [...]
>   
>> +(define-public font-cns11643
>> +  (package
>> +    (name "font-cns11643")
>> +    (version "98.1.20170405")
>> +    (source (origin
>> +              (method url-fetch)
>> +              (uri "http://www.cns11643.gov.tw/AIDB/Open_Data.zip")
>> +              (sha256
>> +               (base32
>> +                "1kyfrwamr8zpdp4qw3z9j3cimhsvr4xd05dmxmfphiknlbr6d455"))))
> I'm getting a different hash when downloading this:
>
> @ build-failed /gnu/store/b4s6pjp7pvwn5028zxm7dn7hxb7akhxn-Open_Data.zip.drv - 1 output path `/gnu/store/kv2291fl8p8r3ic15pp68751qncrv830-Open_Data.zip' should have sha256 hash `1kyfrwamr8zpdp4qw3z9j3cimhsvr4xd05dmxmfphiknlbr6d455', instead has `02kb3bwjrra0k2hlr2p8xswd2y0xs6j8d9vm6yrby734h02a40qf'
>
> Maybe I'm getting a newer version? If this is updated in-place upstream
> very frequently, it would be good to have a way to access older
> downloads. Any ideas?
>
> Minor nit-picks follow, but I've corrected this in my local branch:
I have confirmed that the new hash represents an update to the font. I
have thought about hosting this file, but I'm not sure where I can host
~300MiB. Also, It means I have to bother updating that file frequently.
On the other hand if we use upstream, a failing hash typically means
there is an update. There is unlikely any other Guix user interested in
these fonts at the present. So I'm not sure what to do.
>> +    (outputs '("out" "tw-kai" "tw-sung"))
>> +    (build-system trivial-build-system)
>> +    (native-inputs
>> +     `(("unzip" ,unzip)))
>> +    (arguments
>> +     `(#:modules ((guix build utils))
>> +       #:builder
>> +       (begin
>> +         (use-modules (guix build utils))
>> +         (let* ((font-dir "/share/fonts/truetype/cns11643")
>> +                (out (string-append %output font-dir))
> I've changes this to use (assoc-ref outputs "out") instead of the
> magical %output.
Ok, I don't understand much about those. Are the uses of %outputs (with
the 's') below also problematic?
>> +                (unzip (string-append
>> +                        (assoc-ref %build-inputs "unzip") "/bin/unzip"))
>> +                (tw-kai (string-append
>> +                         (assoc-ref %outputs "tw-kai") font-dir))
>> +                (tw-sung (string-append
>> +                          (assoc-ref %outputs "tw-sung") font-dir)))
>> +           (system* unzip (assoc-ref %build-inputs "source"))
>> +           (chdir "Open_Data/Fonts/")
>> +           (install-file "TW-Kai-98_1.ttf" tw-kai)
>> +           (install-file "TW-Sung-98_1.ttf" tw-sung)
>> +           (install-file "TW-Kai-98_1.ttf" out)
>> +           (install-file "TW-Kai-Ext-B-98_1.ttf" out)
>> +           (install-file "TW-Kai-Plus-98_1.ttf" out)
>> +           (install-file "TW-Sung-98_1.ttf" out)
>> +           (install-file "TW-Sung-Ext-B-98_1.ttf" out)
>> +           (install-file "TW-Sung-Plus-98_1.ttf" out)))))
> And added a #t at the end here, since the return value of "install-file"
> is unspecified.
>
>> [...]
>> +    (license (license:non-copyleft
>> +              "http://data.gov.tw/license"))))
> ...and added a comment about this license being compatible with CC-BY
> 4.0 here.
>
> Thanks a lot for packaging these!

I have attached an updated patch file, and will follow up a with [PATCH
2/2] update