From: suzuki toshiya <mpsuzuki@hiroshima-u.ac.jp>
Date: Fri, 29 Dec 2017 14:04:29 GMT
Subject: Re: [PATCH] git-archive: accepts --owner --group aslike GNU tar.
Message-ID: <5A464B6D.1010102@hiroshima-u.ac.jp>
In-Reply-To: <548d442f05e24e22a1cf4e0074f23f16@OS2PR01MB1147.jpnprd01.prod.outlook.com>

Dear Junio, Ævar

Thank you very much for your reviews, in spite of my many
overlooking of the requirements written in the documents.

To classify various cases, I modified my patch heavily.
It would be posted soon.

I found that some Python scripts are included in the git
repository, but nothing in t/ directories. A helper written
in Python is not welcomed? If so, I would rewrite the
helper in C (perl's standard library does not have a parse
of tarfile). So, please do not review parse-tar-file.py.

Regards,
mpsuzuki

Junio C Hamano wrote:
> suzuki toshiya <mpsuzuki@hiroshima-u.ac.jp> writes:
> 
>> Current tar output by git-archive has always root:root.
> 
> Thanks for a patch.  On top of Ævar's comments...
> 
>> ...
>> * t/t5005-archive-uid-gid.sh: a test script comparing
>>   uid, gid, uname, gname between the options and
>>   generated tar file.
>> ---
> 
> Before the "---" line, we need to get the patch signed off by the
> author (see Documentation/SubmittingPatches).
> 
>> diff --git a/archive-tar.c b/archive-tar.c
>> index c6ed96ee7..8546a6229 100644
>> --- a/archive-tar.c
>> +++ b/archive-tar.c
>> @@ -204,10 +204,10 @@ static void prepare_header(struct archiver_args *args,
>>  	xsnprintf(header->size, sizeof(header->size), "%011lo", S_ISREG(mode) ? size : 0);
>>  	xsnprintf(header->mtime, sizeof(header->mtime), "%011lo", (unsigned long) args->time);
>>  
>> -	xsnprintf(header->uid, sizeof(header->uid), "%07o", 0);
>> -	xsnprintf(header->gid, sizeof(header->gid), "%07o", 0);
>> -	strlcpy(header->uname, "root", sizeof(header->uname));
>> -	strlcpy(header->gname, "root", sizeof(header->gname));
>> +	xsnprintf(header->uid, sizeof(header->uid), "%07o", args->uid);
>> +	xsnprintf(header->gid, sizeof(header->gid), "%07o", args->gid);
>> +	strlcpy(header->uname, args->uname ? args->uname : "root", sizeof(header->uname));
>> +	strlcpy(header->gname, args->gname ? args->gname : "root", sizeof(header->gname));
> 
> Would it be cleaner to make sure aregs->[gu]name is always set
> (i.e. stuff "root" when it is not given)?
> 
>>  	xsnprintf(header->devmajor, sizeof(header->devmajor), "%07o", 0);
>>  	xsnprintf(header->devminor, sizeof(header->devminor), "%07o", 0);
>>  
>> diff --git a/archive.c b/archive.c
>> index 0b7b62af0..db69041f1 100644
>> --- a/archive.c
>> +++ b/archive.c
>> @@ -8,6 +8,7 @@
>>  #include "parse-options.h"
>>  #include "unpack-trees.h"
>>  #include "dir.h"
>> +#include "git-compat-util.h"
> 
> The coding guideline says that "git-compat-util.h" (or one of the
> well-known header that includes it) should be the first file to be
> included, and we already include "cache.h" as the first thing, so
> I do not think you want this addition here.
> 
>> @@ -417,6 +418,57 @@ static void parse_treeish_arg(const char **argv,
>>  	{ OPTION_SET_INT, (s), NULL, (v), NULL, "", \
>>  	  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_HIDDEN, NULL, (p) }
>>  
>> +static void set_args_uname_uid(struct archiver_args *args,
>> +		const char* tar_owner, int set_gid_too)
> 
> The asterisk sticks to the variable name, not the type name, i.e.
> 
> 	const char *tar_owner
> 
>> +{
>> +	if (!args || !tar_owner)
>> +		return;
>> +
>> +	const char* col_pos = strchr(tar_owner, ':');
>> +	struct passwd* pw = NULL;
> 
> Decl after statement.
> 
>> +	if (col_pos) {
>> +		args->uname = xstrndup(tar_owner, col_pos - tar_owner);
>> +		args->uid = atoi(col_pos + 1);
>> +		return;
>> +	}
>> +
>> +	args->uname = xstrndup(tar_owner, strlen(tar_owner));
>> +	pw = getpwnam(tar_owner);
>> +	if (!pw)
>> +		return;
> 
> This means that upon error, the caller gets a half-filled args
> structure and without any indication.
> 
>> diff --git a/t/parse-tar-file.py b/t/parse-tar-file.py
> 
> Hmph.  Do we still use Python around here?
>