git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Merge-Recursive Improvements

From
VSVoltage Spike <voltspike@gmail.com>
Date
Feb 12, 2008, 22:16 UTC
Message-ID
<A21B3CA8-6240-434F-87A9-C6F76DA15265@gmail.com>

I would like to make a series of significant improvements to the merge-recursive mechanism in git, but I was hoping to solicit some early feedback before submitting patches.

First, git is overly zealous at merging differences and two functions added at the same point in a file become intertwined during the merge. A trivial example of this behavior:

   <<<<<<< HEAD:file.txt
   void newfunc1()
   =======
   void newfunc2()
   >>>>>>> merge:file.txt
   {
     int err;
   <<<<<<< HEAD:file.txt
     err = doSomething();
   =======
     err = doSomethingElse();
   >>>>>>> merge:file.txt

Second, git doesn't tell me the original code inside the conflict markers so I almost always resort to "MERGE_HEAD...ORIG_HEAD" and "ORIG_HEAD...MERGE_HEAD" diffs to see what was going on. I could use an external diff tool (yuck), but I would like to modify the conflict markers to resemble those of Perforce:

   >>>>>>> merge-base:file.txt
   Original code.
   ======= HEAD:file.txt
   Head code.
   ======= merge:file.txt
   Merged code.
   <<<<<<<

Third, git doesn't appear to have any sense of context when performing a merge. Another contrived example which wouldn't be flagged as a merge conflict:

   ptr = malloc(len); // Added in HEAD.
   init();            // Included in merge-base.
   ptr = malloc(len); // Added in "merge".

Fourth, git doesn't provide a mechanism for merges to ignore whitespace changes.

I resolved issues the first and the fourth through the introduction of new configuration variables and trivial modifications to the manner in which we call xdl_merge. I suspect the second and third issue may also be simple to solve but would require that I modify libxdiff directly.

Are these changes something other people might be interested in? (It seems odd that nobody is complaining about these really irritating flaws.) Should I concern myself with writing a custom merge driver rather than modify core behavior (even if the change is configurable)? If I should focus on an external driver, under what circumstances would merge.*.recursive come into play (i.e., when do I have to worry about poor behavior for an "internal merge")?

Thank you in advance for the feedback.
Next: Stefan Monnier
Message 1 of 23 in “Merge-Recursive Improvements”
  1. Voltage SpikeFeb 12, 2008
  2. Stefan MonnierFeb 12, 2008
  3. Junio C HamanoFeb 12, 2008
  4. Linus TorvaldsFeb 12, 2008
  5. Johannes SchindelinFeb 13, 2008
  6. xdl_merge(): introduce XDL_MERGE_ZEALOUS_ALNUMJohannes Schindelin, Feb 13, 2008
  7. Junio C HamanoFeb 13, 2008
  8. Johannes SchindelinFeb 13, 2008
  9. Junio C HamanoFeb 15, 2008
  10. Linus TorvaldsFeb 15, 2008
  11. Johannes SchindelinFeb 15, 2008
  12. Johannes SchindelinFeb 17, 2008
  13. 1/2 xdl_merge(): make XDL_MERGE_ZEALOUS output simplerJohannes Schindelin, Feb 17, 2008
  14. 2/2 xdl_merge(): introduce XDL_MERGE_ZEALOUS_ALNUMJohannes Schindelin, Feb 17, 2008
  15. Junio C HamanoFeb 18, 2008
  16. Johannes SchindelinFeb 18, 2008
  17. Linus TorvaldsFeb 13, 2008
  18. Johannes SchindelinFeb 13, 2008
  19. Johannes SixtFeb 13, 2008
  20. Steffen ProhaskaFeb 13, 2008
  21. Voltage SpikeFeb 13, 2008
  22. Johannes SixtFeb 13, 2008
  23. Junio C HamanoFeb 15, 2008

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.