Re: [meta-multimedia][PATCH] flite: add recipe


Khem Raj
 

On Tue, May 31, 2022 at 8:35 AM Scott Murray <scott.murray@...> wrote:

On Tue, 31 May 2022, Samuli Piippo wrote:

On Mon, 30 May 2022 at 17:43, Scott Murray <scott.murray@...>
wrote:

On Mon, 30 May 2022, Samuli Piippo wrote:

CMU Flite (festival-lite) is a small, fast run-time open source
text to speech synthesis engine developed at CMU and primarily
designed for small embedded machines.

Signed-off-by: Samuli Piippo <samuli.piippo@...>
---
.../recipes-multimedia/flite/flite_2.2.bb | 19 +++++++++++++++++++
1 file changed, 19 insertions(+)
create mode 100644 meta-multimedia/recipes-multimedia/flite/
flite_2.2.bb

diff --git a/meta-multimedia/recipes-multimedia/flite/flite_2.2.bb
b/meta-multimedia/recipes-multimedia/flite/flite_2.2.bb
new file mode 100644
index 000000000..9ab9c0208
--- /dev/null
+++ b/meta-multimedia/recipes-multimedia/flite/flite_2.2.bb
@@ -0,0 +1,19 @@
+SUMMARY = "Flite: a small run-time speech synthesis engine"
+HOMEPAGE = "http://cmuflite.org"
+SECTION = "libs/multimedia"
+LICENSE = "BSD-3-Clause"
+LIC_FILES_CHKSUM = "file://COPYING;md5=b3b732d1349633a53e69356607fd2d6c"
+
+inherit autotools-brokensep
+
+SRC_URI = "git://
github.com/festvox/flite.git;protocol=https;branch=master"
+
+SRCREV = "e9e2e37c329dbe98bfeb27a1828ef9a71fa84f88"
[snip]

There's no 2.3 tag, but the README.md and Makefile imply that 2.3 was
released in March 2022 (see commit 530060de), and the current voice files
on http://festvox.org/flite/packed/ are "flite-2.3". So perhaps this
should be building that instead of the ~2 years old 2.2?
Hard to say if there's a specific revision that was meant to be the
release, but I could update this to 2.3 and just use the latest revision?
I'd be in favor of a _git recipe that set PV = "2.3+git${SRCPV}", but Khem
might have a different opinion.
yeah if there is no tag then this is a better approach.


+
+S = "${WORKDIR}/git"
+
+EXTRA_OECONF += "--enable-shared"
+
+do_install:append() {
+ chown -R root:root ${D}${libdir}/*
This seems a bit odd, I'm curious what prompted adding it?
The installed files had local user and group IDs that caused host
contamination error.
Hrm, that suggests upstream have a broken Makefile using "cp -p", it could
be worthwhile to poke them about it. We have had a recipe for an older
version in AGL for quite a while and didn't see it, so it seems like
something changed.
right, its perhaps in install target of makefile where it maybe the
better place to fix
it and also upstream it.


Scott



Join openembedded-devel@lists.openembedded.org to automatically receive all group messages.