Received: from out.migadu.com (out.migadu.com [91.121.223.63]) by nld3-dev1.alpinelinux.org (Postfix) with ESMTPS id AF30B78107B for <~alpine/apk-tools@lists.alpinelinux.org>; Wed, 19 Feb 2020 00:58:32 +0000 (UTC) Received: (Migadu outbound); Wed, 19 Feb 2020 00:58:31 +0000 Authentication-Results: out.migadu.com; auth=pass (plain) Received: from wms0-eu-central.migadu.com (wms0-eu-central.migadu.com [139.162.159.86]) by out.migadu.com (Haraka/2.8.16) with ESMTPSA id 05BCD26B-68DC-433D-A26D-3669A3C045D6.1 envelope-from (authenticated bits=0) (version=TLSv1/SSLv3 cipher=ECDHE-RSA-AES256-GCM-SHA384 verify=FAIL); Wed, 19 Feb 2020 00:58:31 +0000 MIME-Version: 1.0 Date: Wed, 19 Feb 2020 00:58:31 +0000 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable X-Mailer: RainLoop/1.12.1 From: "Ariadne Conill" Message-ID: <340209c209104e20dd01b7b6aa71251c@dereferenced.org> Subject: Re: APK Package Name Issue on armel port To: "Timo Teras" , "Carl Chave" Cc: ~alpine/apk-tools@lists.alpinelinux.org In-Reply-To: <20200217123334.031047d1@vostro.wlan> References: <20200217123334.031047d1@vostro.wlan> <20200214105215.6873e402@vostro.wlan> <62f21c96-ff43-83e4-5b7f-2e65e6d72729@adelielinux.org> <20200217111200.08405b9c@vostro.wlan> DKIM-Signature: v=1;a=rsa-sha256;bh=0arpUVs67FjIDO1eyN2bt/i7rRq3w/wiOMaXp2f+KqY=;c=relaxed/simple;d=dereferenced.org;h=from:subject:date:to;s=default;b=FsogCw4s2IiYpAy0rZzWl1EvQN0zWW1ZJfYuFTz8hJVuTNdGpQEKWIdPkdTCIi/0v9S/W/YqGzIc7Mt+J20n+BRnhgcYtsj9BpFoTeo7klTaFqmQWWrQI6y9scpzg1pN4uNgBOCiPZAR7b5F7ww6nxxwsHablqT4IF003saAcg0= Hello,=0A=0AFebruary 17, 2020 4:33 AM, "Timo Teras" w= rote:=0A=0A> On Mon, 17 Feb 2020 11:12:00 +0200=0A> Timo Teras wrote:=0A> =0A>> Reviewing the code in apk, it seems that the m= urmur hashing code is=0A>> not properly accounting for alignment. So the = symptoms do match. The=0A>> hash lookup does not work (well, works random= ly based on few things),=0A>> but when enumerating all packages (by using= the wildcard in lookups)=0A>> it'll show everything.=0A> =0A> Seems this= is a common complaint about original murmur3 and there's=0A> been other = projects affected with this too.=0A=0AI wonder if we may be better off ad= opting a simpler hash function=0Afor our hashtables, such as FNV-1? I ha= ve had much success over=0Athe years using FNV hashes in various projects= in a similar role=0Aas what apk uses murmur3 for.=0A=0AAriadne