> On Apr 23, 2019, at 2:45 PM, Matthew Flatt <mfl...@cs.utah.edu> wrote:
>
> At Tue, 23 Apr 2019 23:39:04 +0200, Ryan Culpepper wrote:
>> It looks like natipkg Racket includes libharfbuzz.so.0, not
>> libharfbuzz.so.1 (see [1]). So try changing the unix line to
>>
>> [(unix) (ffi-lib "libharfbuzz" '("0" ""))]
>>
>> instead.
>
> Right.
>
> Just to elaborate a little, there is no "libharfbuzz.so.1", so by using
> '("1" "") you end up picking up "libharfbuzz.so" from the OS instead of
> the "libharfbuzz.so.0" natipkg build, and the pkg-build OS is old.
Yes, that advice cleared the error on the Racket package server.
But now it fails on Travis with a similar error (I know you are not Travis
support, but let's pretend Travis represents an arbitrary Linux user who might
use this package):
ffi-obj: couldn't get "hb_buffer_add_codepoints" from "libharfbuzz.so.0"
(/usr/lib/x86_64-linux-gnu/libharfbuzz.so.0: undefined symbol:
hb_buffer_add_codepoints)
IIUC, even though I'm now invoking "libharfbuzz.so.0" explicitly, in this case
the Travis server OS has its own "libharfbuzz.so.0" that is conflicting with
the one furnished by Racket.
So is there a way to be even more specific about which "libharfbuzz.so.0" I
load with `define-runtime-lib`?
--
You received this message because you are subscribed to the Google Groups
"Racket Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to racket-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
