Talk:Atomic commit

More information / clarification requested

Please explicitly describe the atomic commit "problem" illustrated by the Two Generals' problem.

In other words, please describe the connection between atomic commits and the Two Generals' problem.

I feel the connection is not obvious.

• Seconded. When I hit the sentence "It has been proven that no algorithm can solve the problem" I thought "Huh!? What problem? There was no previous mention of a problem.". --Irrevenant [ talk ] 07:06, 11 March 2009 (UTC)

Merge

Recommend merging this and other SQL transaction statements into a subsection of Database transactions article and redirecting these statements there. Comments? SqlPac 05:29, 17 May 2007 (UTC)

Definitely No. This is a sufficiently large subject. However the article requires a major rewrite, and merging with atomic commitment which does not exist yet Comps 11:17, 24 May 2007 (UTC)

Sufficiently large? There are only 6 sentences here. Is this subject really that large that it requires multiple articles to cover it? There are articles on Atomic commit, two-phase and three-phase commit protocol, Commit (data management), rollback (data management), cascading rollback, Atomicity, Atomic operation, Transaction processing. Most of these articles are a couple of paragraphs at most... I think there might be one good article's worth of information spread among all these stubs. SqlPac 19:33, 24 May 2007 (UTC)

I do not agree. Enough material exists to write about generic atomic commitment. New AC protocols with specific properties have been proposed beyond 2PC, 3PC, etc. This article should be a center and directory for all of them. And we shall see more... Comps 04:33, 22 June 2007 (UTC)

I also vote strongly against merging this article with Database transactions. Atomic commit is not specific to DBs. It is a generic distributed computing problem, and there are many algorithms that solve it. Databases are just one scenario where an atomic commit must be performed. Jaksa 18:22, 4 July 2007 (UTC)

I disagree as well. For example, atomic commits exist in revision control (i. e. not a database transaction). -Anon

I agree with Jaksa but maybe Atomicity and Atomic commit could go under Atomic operation. If so, the link from Comparison_of_revision_control_software to Atomic commit should change also --CoreTechX 10:53, 31 October 2007 (UTC)

no —Preceding unsigned comment added by 195.195.47.18 (talk) 10:59, 16 April 2008 (UTC)

Split page

This page talks about 3 or so separate topics without successfully relating them together. It switches between them without adequate introduction: - Atomic commits as in ACID, committed to permanent storage in a database. - Atomic commits in a distributed/networked data store and atomic commit protocols (and two generals problem). - Commits in a source code repository that are small and focused on a single feature. I don't even think this should be referred to as "atomic commits" - is that really an established listed term for it? Works needs to be done to separate and introduce the different concepts and maybe they need to be on separate pages. mmj (talk) 11:35, 27 November 2017 (UTC) mmj (talk) 11:35, 27 November 2017 (UTC)