From: Philip Oakley Date: Tue, 30 May 2017 23:04:46 GMT Subject: Re: mergetool: what to do about deleting precious files? Message-ID: In-Reply-To: From: "Junio C Hamano" Thanks for the replies. Let's see if I've got it... > "Philip Oakley" writes: > >> If I now understand correctly, the merge process flow is: >> >> * canonicalise content (eol, smudge-clean, $id, renormalise, etc) >> * diff the content (internal, or GIT_EXTERNAL_DIFF) >> * apply the diff >> * if conflicts, only then use merge-driver/tool >> >> Would that be a correct interpretation? > > Not quite. There are a lot more going on before any of those steps: > > * Find the common ancestor commit (which could be many). IIUC Git selects one of them, rather than all if there are many (which then may not be the optimum) > > * Walk the three trees (the common ancestor's, ours and theirs) in > parallel, noticing what happened to each path. Depending on what > happened to the path in each branch, the merge may or may not > "conflict" (e.g. when both sides added exactly the same contents > to the same path, they are not counted as conflicting. when we > removed while they modified, they show as conflicting). I'm assuming here that this is the sha-oid comparison, and then checking the tree/blob names that match them. (the top tree not having a name). So here "conflict free" is that the sha-oids match. Also, I thnk this is saying that added or removed trees or blobs are in some sense are 'conflict free' (though still subject to rename/move detection etc). An added file/blob would be conflict free for merging into it's tree, yes? IIUC, the comparison is therefore using the in-repo sha-oids; unless --renormalise was given which will do a smudge-clean washing cycle and recomute fresh canonical sha-oids for the comparison (rather than doing it later). > > * For paths that are conflicting, feed the canonicalized content of > the versions from common, ours and theirs to the file-level merge > driver. So this is where any .gitattibutes settings come in, or is the merge driver after the diff step? (which could also be a user diff?) > The builtin file-level merge driver takes two xdiff (one > between ancestor and ours, the other between ancestore and > theirs) and reconciles them to produce the result. But that is > irrelevant in the context of "custom merge driver"; the builtin > one is skipped altogether and the custom contents merge driver > the user specified via the attributes is used instead. > > Notice that the second step above has no customization knobs. Any > path the second step deems not to conflict is "merged cleanly" > without even triggering the "oops, ours and theirs did conflicting > changes, to the content; let's see how the final content should look > like" (aka the third step). This is *not* because "Git knows the > best"; it is merely that nobody felt the need for a mechanism to > allow customizing the second step. > > And that is why I said you need a new customization mechanism if you > want to affect the outcome of the scenario that started this thread. Ok, I think I can see how I was confused between the "tree merge" (oid conflict detection) and the more usual (to users) "file merge" (line by line, etc.). I wasn't sure where to find that as someone relatively new to Git. Thanks for the explanations. -- Philip --- This email has been checked for viruses by AVG. http://www.avg.com