From: Junio C Hamano <gitster@pobox.com>
To: Michael Haggerty <mhagger@alum.mit.edu>
Cc: Jiang Xin <worldhello.net@gmail.com>, Git List <git@vger.kernel.org>
Subject: Re: [PATCH v13 02/15] path.c: refactor relative_path(), not only strip prefix
Date: Wed, 22 May 2013 09:23:33 -0700 [thread overview]
Message-ID: <7v38tfkmpm.fsf@alter.siamese.dyndns.org> (raw)
In-Reply-To: <519C7CA1.2060402@alum.mit.edu> (Michael Haggerty's message of "Wed, 22 May 2013 10:06:57 +0200")
Michael Haggerty <mhagger@alum.mit.edu> writes:
>> Different results for relative_path() before and after this refactor:
>>
>> abs path base path relative (original) relative (refactor)
>> ======== ========= =================== ===================
>> /a/b/c/ /a/b c/ c/
>> /a/b//c/ //a///b/ c/ c/
>> /a/b /a/b . ./
>> /a/b/ /a/b . ./
>> /a /a/b/ /a ../
>> / /a/b/ / ../../
>> /a/c /a/b/ /a/c ../c
>> /a/b (empty) /a/b /a/b
>> /a/b (null) /a/b /a/b
>> (empty) /a/b (empty) ./
>> (null) (empty) (null) ./
>> (null) /a/b (segfault) ./
>
> The old and new versions both seem to be (differently) inconsistent
> about when the output has a trailing slash. What is the rule?
That is a good point. At least adding / at the end of "." seems
unneeded, given that the output in some cases have no trailing
slash, forcing a caller who wanted to get a directory to append a
trailing path components to it to check if it needs to add one
before doing so. Always adding a slash / to the output may sound
consistent, but it is not quite; e.g. "/a/c based on /a/b is ../c"
case may be referring to a non directory /a/c and ensuring a
trailing slash to produce ../c/ is actively wrong.
"The caller knows" rule might work (I am thinking aloud, without
looking at existing callers to see what would break, only to see if
a consistent and simple-to-explain rule is possible).
When the caller asks to turn /a/b relative to /a/b (or /a/b/,
/a//b/./), then we do not know (or we do not want to know) if the
caller means it to be a directory with the intention of appending
something after it, so just return ".". When the caller asks to
turn /a/b/ relative to the same base, return "./" with the trailing
slash. Remember if the "abs path" side had a trailing slash,
normalize both input (turning "/./" into "/" and "foo/bar/../" into
foo/", squashing multiple slashes into one, etc.) and then strip the
trailing slash from them, do the usual "comparison and replacement
of leading path components with series of ../" and then append a
slash if the original had one, or something?
>> index 04ff..0174d 100644
>> --- a/path.c
>> +++ b/path.c
>> @@ -441,42 +441,104 @@ int adjust_shared_perm(const char *path)
>> return 0;
>> }
>>
>> -const char *relative_path(const char *abs, const char *base)
>> +/*
>> + * Give path as relative to prefix.
>> + *
>> + * The strbuf may or may not be used, so do not assume it contains the
>> + * returned path.
>> + */
>> +const char *relative_path(const char *abs, const char *base,
>> + struct strbuf *sb)
>
> Thanks for adding documentation. But I think it could be improved:
>
> * The comment refers to "path" and "prefix" but the function parameters
> are "abs" and "base". I suggest making them agree.
>
> * Who owns the memory pointed to by the return value?
>
> * The comment says that "the strbuf may or may not be used". So why is
> it part of the interface? (I suppose it is because the strbuf might be
> given ownership of the returned memory if it has to be allocated.)
> Would it be more straightforward to *always* return the result in the
> strbuf?
This comes from the original in quote.c, I think. The caller
supplies a strbuf as a scratchpad area and releasing it is the
caller's responsibility. If the function does not need any
scratchpad area (i.e. the answer is a substring of the input), the
result may point into the abs.
So the calling convention is:
struct strbuf scratch = STRBUF_INIT;
const char *result = relative(abs, base, &scratch);
use(result);
strbuf_release(&scratch);
and the lifetime rule is "consume it before either abs or scratch
is changed".
next prev parent reply other threads:[~2013-05-22 16:23 UTC|newest]
Thread overview: 54+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-05-17 9:20 [PATCH v11 00/15] *** Interactive git-clean *** Jiang Xin
2013-05-17 9:20 ` [PATCH v11 01/15] path.c: refactor relative_path(), not only strip prefix Jiang Xin
2013-05-17 9:20 ` [PATCH v11 02/15] test: test relative_path through test-path-utils Jiang Xin
2013-05-17 9:20 ` [PATCH v11 03/15] quote.c: remove path_relative, use relative_path instead Jiang Xin
2013-05-17 9:20 ` [PATCH v11 04/15] Refactor quote_path_relative, remove unused params Jiang Xin
2013-05-17 9:20 ` [PATCH v11 05/15] Refactor write_name_quoted_relative, " Jiang Xin
2013-05-17 9:20 ` [PATCH v11 06/15] git-clean: refactor git-clean into two phases Jiang Xin
2013-05-17 9:20 ` [PATCH v11 07/15] git-clean: add support for -i/--interactive Jiang Xin
2013-05-17 9:20 ` [PATCH v11 08/15] git-clean: show items of del_list in columns Jiang Xin
2013-05-17 9:20 ` [PATCH v11 09/15] git-clean: add colors to interactive git-clean Jiang Xin
2013-05-17 9:20 ` [PATCH v11 10/15] git-clean: use a git-add-interactive compatible UI Jiang Xin
2013-05-17 9:20 ` [PATCH v11 11/15] git-clean: add filter by pattern interactive action Jiang Xin
2013-05-17 9:20 ` [PATCH v11 12/15] git-clean: add select by numbers " Jiang Xin
2013-05-17 9:20 ` [PATCH v11 13/15] git-clean: add ask each " Jiang Xin
2013-05-17 9:20 ` [PATCH v11 14/15] git-clean: add documentation for interactive git-clean Jiang Xin
2013-05-17 9:20 ` [PATCH v11 15/15] test: add t7301 for git-clean--interactive Jiang Xin
2013-05-18 3:18 ` [PATCH v12 00/15] Interactive git-clean Jiang Xin
2013-05-20 22:48 ` Junio C Hamano
2013-05-18 3:18 ` [PATCH v12 01/15] test: add test cases for relative_path Jiang Xin
2013-05-21 20:37 ` Junio C Hamano
2013-05-22 1:40 ` [PATCH v13 00/15] interactive git-clean Jiang Xin
2013-05-22 1:40 ` [PATCH v13 01/15] test: add test cases for relative_path Jiang Xin
2013-05-22 1:40 ` [PATCH v13 02/15] path.c: refactor relative_path(), not only strip prefix Jiang Xin
2013-05-22 8:06 ` Michael Haggerty
2013-05-22 16:23 ` Junio C Hamano [this message]
2013-05-26 0:21 ` Jiang Xin
2013-05-30 6:56 ` Jiang Xin
2013-05-22 1:40 ` [PATCH v13 03/15] quote.c: remove path_relative, use relative_path instead Jiang Xin
2013-05-22 1:40 ` [PATCH v13 04/15] Refactor quote_path_relative, remove unused params Jiang Xin
2013-05-22 1:40 ` [PATCH v13 05/15] Refactor write_name_quoted_relative, " Jiang Xin
2013-05-22 1:40 ` [PATCH v13 06/15] git-clean: refactor git-clean into two phases Jiang Xin
2013-05-22 1:40 ` [PATCH v13 07/15] git-clean: add support for -i/--interactive Jiang Xin
2013-05-22 1:40 ` [PATCH v13 08/15] git-clean: show items of del_list in columns Jiang Xin
2013-05-22 1:40 ` [PATCH v13 09/15] git-clean: add colors to interactive git-clean Jiang Xin
2013-05-22 1:40 ` [PATCH v13 10/15] git-clean: use a git-add-interactive compatible UI Jiang Xin
2013-05-22 1:40 ` [PATCH v13 11/15] git-clean: add filter by pattern interactive action Jiang Xin
2013-05-22 1:40 ` [PATCH v13 12/15] git-clean: add select by numbers " Jiang Xin
2013-05-22 1:40 ` [PATCH v13 13/15] git-clean: add ask each " Jiang Xin
2013-05-22 1:40 ` [PATCH v13 14/15] git-clean: add documentation for interactive git-clean Jiang Xin
2013-05-22 1:40 ` [PATCH v13 15/15] test: add t7301 for git-clean--interactive Jiang Xin
2013-05-18 3:18 ` [PATCH v12 02/15] path.c: refactor relative_path(), not only strip prefix Jiang Xin
2013-05-18 3:18 ` [PATCH v12 03/15] quote.c: remove path_relative, use relative_path instead Jiang Xin
2013-05-18 3:18 ` [PATCH v12 04/15] Refactor quote_path_relative, remove unused params Jiang Xin
2013-05-18 3:18 ` [PATCH v12 05/15] Refactor write_name_quoted_relative, " Jiang Xin
2013-05-18 3:18 ` [PATCH v12 06/15] git-clean: refactor git-clean into two phases Jiang Xin
2013-05-18 3:18 ` [PATCH v12 07/15] git-clean: add support for -i/--interactive Jiang Xin
2013-05-18 3:19 ` [PATCH v12 08/15] git-clean: show items of del_list in columns Jiang Xin
2013-05-18 3:19 ` [PATCH v12 09/15] git-clean: add colors to interactive git-clean Jiang Xin
2013-05-18 3:19 ` [PATCH v12 10/15] git-clean: use a git-add-interactive compatible UI Jiang Xin
2013-05-18 3:19 ` [PATCH v12 11/15] git-clean: add filter by pattern interactive action Jiang Xin
2013-05-18 3:19 ` [PATCH v12 12/15] git-clean: add select by numbers " Jiang Xin
2013-05-18 3:19 ` [PATCH v12 13/15] git-clean: add ask each " Jiang Xin
2013-05-18 3:19 ` [PATCH v12 14/15] git-clean: add documentation for interactive git-clean Jiang Xin
2013-05-18 3:19 ` [PATCH v12 15/15] test: add t7301 for git-clean--interactive Jiang Xin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=7v38tfkmpm.fsf@alter.siamese.dyndns.org \
--to=gitster@pobox.com \
--cc=git@vger.kernel.org \
--cc=mhagger@alum.mit.edu \
--cc=worldhello.net@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).