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

Re: [PATCH 4/9] read_early_config(): optionally return the worktree's top-level directory

From
BWBrandon Williams <bmwill@google.com>
Date
Jun 8, 2017, 14:46 UTC
Message-ID
<20170608144628.GH110638@google.com>
In-Reply-To
<alpine.DEB.2.21.1.1706081204330.171564@virtualbox>
On 06/08, Johannes Schindelin wrote:
Show 45 quoted lines
> Hi Brandon,
> 
> On Wed, 7 Jun 2017, Brandon Williams wrote:
> 
> > On 06/07, Johannes Schindelin wrote:
> > > @@ -1668,7 +1668,7 @@ void read_early_config(config_fn_t cb, void *data)
> > >  	 * notably, the current working directory is still the same after the
> > >  	 * call).
> > >  	 */
> > > -	else if (discover_git_directory(&buf))
> > > +	else if (discover_git_directory(&buf, worktree_dir))
> > >  		opts.git_dir = buf.buf;
> > 
> > It feels kind of weird to get back worktree info after calling
> > read_early_config but I understand why you need to get it.
> 
> Yeah. It is awkward. Required for backwards-compatibility, though (and it
> is hard to explain *when* it is needed, and even harder under what
> circumstances it is *not* needed even if there is a worktre).
> 
> > One thing to consider is the 'if' statement not shown here since you
> > aren't adding any worktree information if that branch is taken.
> 
> Right. For lurkers, that `if` statement reads thusly:
> 
> 	if (have_git_dir())
> 		opts.git_dir = get_git_dir();
> 
> > Maybe we can drop the first if statement all together as
> > 'read_early_config' is used before setup is run and it should really
> > only be triggered when setup has been run.
> 
> The `read_early_config()` function is *sometimes* used *after* setup has
> run. Look at `run_builtin()` in git.c:
> 
> 	if (p->option & RUN_SETUP)
> 		prefix = setup_git_directory();
> 	else if (p->option & RUN_SETUP_GENTLY) {
> 		int nongit_ok;
> 		prefix = setup_git_directory_gently(&nongit_ok);
> 	}
> 
> 	if (use_pager == -1 && p->option & (RUN_SETUP | RUN_SETUP_GENTLY))
> 		use_pager = check_pager_config(p->cmd);
> 
Ah, ok so my comment was incorrect, thanks for pointing this out.
Show 14 quoted lines
> For builtins having either the RUN_SETUP or the RUN_SETUP_GENTLY flag, we
> do not need to re-discover the .git/ directory at all when checking the
> pager config.
> 
> Back to the worktree_dir variable.
> 
> I think part of the confusion here is that it may be left alone even when
> there is a worktree. For example, if we are already in the top-level
> directory. Or if the worktree somehow points to a different directory than
> the one containing the .git/ directory.
> 
> Therefore, I renamed this variable to `cdup_dir` to reflect the fact that
> it is only touched if Git determines that it is in a subdirectory of the
> directory containing the .git/ directory.

Ok, maybe I'm just not following but just from reading the variable name I can't really understand what 'cdup_dir' means.

> 
> Ciao,
> Dscho
-- 
Brandon Williams
Previous: Johannes SchindelinNext: Johannes Schindelin
Message 14 of 34 in “Avoid problem where git_dir is set after alias expansion”
  1. 0/9 Avoid problem where git_dir is set after alias expansionJohannes Schindelin, Jun 7, 2017
  2. 1/9 discover_git_directory(): avoid setting invalid git_dirJohannes Schindelin, Jun 7, 2017
  3. Brandon WilliamsJun 7, 2017
  4. Johannes SchindelinJun 8, 2017
  5. 2/9 config: report correct line number upon errorJohannes Schindelin, Jun 7, 2017
  6. Junio C HamanoJun 9, 2017
  7. 3/9 help: use early config when autocorrecting aliasesJohannes Schindelin, Jun 7, 2017
  8. Brandon WilliamsJun 7, 2017
  9. Johannes SchindelinJun 8, 2017
  10. Junio C HamanoJun 9, 2017
  11. 4/9 read_early_config(): optionally return the worktree's top-level directoryJohannes Schindelin, Jun 7, 2017
  12. Brandon WilliamsJun 7, 2017
  13. Johannes SchindelinJun 8, 2017
  14. Brandon WilliamsJun 8, 2017
  15. Johannes SchindelinJun 8, 2017
  16. Brandon WilliamsJun 8, 2017
  17. Johannes SchindelinJun 8, 2017
  18. Brandon WilliamsJun 8, 2017
  19. 7/9 alias_lookup(): optionally return top-level directoryJohannes Schindelin, Jun 7, 2017
  20. 6/9 t7006: demonstrate a problem with aliases in subdirectoriesJohannes Schindelin, Jun 7, 2017
  21. 5/9 t1308: relax the test verifying that empty alias values are disallowedJohannes Schindelin, Jun 7, 2017
  22. Brandon WilliamsJun 7, 2017
  23. Johannes SchindelinJun 8, 2017
  24. Brandon WilliamsJun 8, 2017
  25. Junio C HamanoJun 10, 2017
  26. 8/9 Use the early config machinery to expand aliasesJohannes Schindelin, Jun 7, 2017
  27. Brandon WilliamsJun 7, 2017
  28. Johannes SchindelinJun 8, 2017
  29. Brandon WilliamsJun 8, 2017
  30. Junio C HamanoJun 10, 2017
  31. Johannes SchindelinJun 7, 2017
  32. Brandon WilliamsJun 7, 2017
  33. Johannes SchindelinJun 8, 2017
  34. Brandon WilliamsJun 8, 2017

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.