From: Peter Korsgaard Date: Tue, 07 Oct 2014 23:12:03 +0200 Subject: [Buildroot] [PATCH 1/3] package/libva: Bump version to 1.4.0 In-Reply-To: <20141007210631.GA27580@free.fr> (Yann E. MORIN's message of "Tue, 7 Oct 2014 23:06:31 +0200") References: <1412623692-3322-1-git-send-email-bernd.kuhls@t-online.de> <87ppe3fwnn.fsf@dell.be.48ers.dk> <20141007210631.GA27580@free.fr> Message-ID: <878ukrfuu4.fsf@dell.be.48ers.dk> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net >>>>> "Yann" == Yann E MORIN writes: Hi, > ... there are a few other good reasons we accept hashes: > - it ensures a broken download is detected, so the user quickly knows > that the tarball is broken because of the download; sometimes, > upstreams breaks their distributions (e.g. the recent sourceforge > breakage...). > - non-compliant downloads are removed (rm -f) so they are not > accidentally used in another context (e.g. I do share my BR2_DL_DIR > with other stuff). > - consequently, it helps the autobuilders prune their failed > downloads. > But in the end there is no clear policy, except: > - we *do* want hashes for security-related packages; > - hashes for other packages are a nice bonus. Agreed. It is in no way required, but nice - Especially if upstream posts hashes. -- Bye, Peter Korsgaard