Re: [PATCH v3 2/9] Disallow diffopt.close_file when using the log_tree machinery

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Johannes Schindelin <johannes.schindelin@xxxxxx> writes:

> We are about to teach the log_tree machinery to reuse the diffopt.file
> setting to output to a file stream different from stdout.
>
> This means that builtin am can no longer ask the diff machinery to
> close the file when actually calling the log_tree machinery (which
> wants to flush the very same file stream that would then already be
> closed).

Sorry for being slow, but I am not sure why the first paragraph has
to mean the second paragraph.  This existing caller opens a new
stream, sets .fp to it, and expects that the log_tree_commit() to
close it if told by setting .close_file to true, all of which sounds
sensible.

If a codepath wants to use the same stream for two or more calls to
log_tree by pointing the stream with .fp, it would be of course a
problem for the caller to set .close_file to true in its first call,
as .fp will be closed and no longer usable for second and subsequent
call, and that would be a bug, but for a single-shot call it feels
entirely a sensible request to make, no?

Obviously you have looked at the codepaths involved a lot longer
than I did, and I do not doubt your conclusion, but I cannot quite
convince myself with the above explanation.

The option parser of "git diff" family sets ->close_file to true
when the --output option is given.

Wouldn't this patch break "git log --output=foo -3"?

> To stave off similar problems in the future, report it as a bug if
> log_tree_commit() is called with a non-zero diffopt.close_file.
>
> Signed-off-by: Johannes Schindelin <johannes.schindelin@xxxxxx>
> ---
>  builtin/am.c | 6 ++++--
>  log-tree.c   | 3 +++
>  2 files changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/builtin/am.c b/builtin/am.c
> index 0e28a62..47d78aa 100644
> --- a/builtin/am.c
> +++ b/builtin/am.c
> @@ -1437,6 +1437,7 @@ static int write_commit_patch(const struct am_state *state, struct commit *commi
>  {
>  	struct rev_info rev_info;
>  	FILE *fp;
> +	int res;
>  
>  	fp = fopen(am_path(state, "patch"), "w");
>  	if (!fp)
> @@ -1453,10 +1454,11 @@ static int write_commit_patch(const struct am_state *state, struct commit *commi
>  	DIFF_OPT_SET(&rev_info.diffopt, FULL_INDEX);
>  	rev_info.diffopt.use_color = 0;
>  	rev_info.diffopt.file = fp;
> -	rev_info.diffopt.close_file = 1;
>  	add_pending_object(&rev_info, &commit->object, "");
>  	diff_setup_done(&rev_info.diffopt);
> -	return log_tree_commit(&rev_info, commit);
> +	res = log_tree_commit(&rev_info, commit);
> +	fclose(fp);
> +	return res;
>  }
>  
>  /**
> diff --git a/log-tree.c b/log-tree.c
> index 78a5381..dc0180d 100644
> --- a/log-tree.c
> +++ b/log-tree.c
> @@ -864,6 +864,9 @@ int log_tree_commit(struct rev_info *opt, struct commit *commit)
>  	struct log_info log;
>  	int shown;
>  
> +	if (opt->diffopt.close_file)
> +		die("BUG: close_file is incompatible with log_tree_commit()");
> +
>  	log.commit = commit;
>  	log.parent = NULL;
>  	opt->loginfo = &log;
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html



[Index of Archives]     [Linux Kernel Development]     [Gcc Help]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [V4L]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]     [Fedora Users]