Question About 'ar' Output

Question About 'ar' Output

[Jon Forrest]

(GNU ar version 2.37-10.fc35)

As a learning exercise, I'm trying to duplicate some of what
the 'ar' command does. (There's nothing wrong with 'ar').

I have a .a library that ar shows has the following content:

% ar t libctf.a
libctf_la-ctf-archive.o
libctf_la-ctf-dump.o
libctf_la-ctf-create.o
libctf_la-ctf-decl.o
libctf_la-ctf-error.o
libctf_la-ctf-hash.o
libctf_la-ctf-labels.o
[rest omitted]

When I run my test program I see
/
//
/0
/25
/47
/71
/93
[rest omitted]

When I look at my .a library using 'od' I can see the names starting
with '/' characters. I can also see the names starting with 'libctf'.

I don't understand what the slash names are, nor how they're mapped to
the real names. I tried looking at the 'ar' source code but I didn't
see anything that jumped out at me.

What am I missing?

Cordially,
Jon Forrest

Re: Question About 'ar' Output

[Jan Beulich]

On 26.04.2022 05:46, Jon Forrest via Binutils wrote:
> (GNU ar version 2.37-10.fc35)
> 
> As a learning exercise, I'm trying to duplicate some of what
> the 'ar' command does. (There's nothing wrong with 'ar').
> 
> I have a .a library that ar shows has the following content:
> 
> % ar t libctf.a
> libctf_la-ctf-archive.o
> libctf_la-ctf-dump.o
> libctf_la-ctf-create.o
> libctf_la-ctf-decl.o
> libctf_la-ctf-error.o
> libctf_la-ctf-hash.o
> libctf_la-ctf-labels.o
> [rest omitted]
> 
> When I run my test program I see
> /
> //
> /0
> /25
> /47
> /71
> /93
> [rest omitted]
> 
> When I look at my .a library using 'od' I can see the names starting
> with '/' characters. I can also see the names starting with 'libctf'.
> 
> I don't understand what the slash names are, nor how they're mapped to
> the real names. I tried looking at the 'ar' source code but I didn't
> see anything that jumped out at me.

Member names, when they're longer than what fits in the designated
member header field, can use an escape mechanism by indexing into the
special "//" member using the "/<offset>" notation.

Jan

Re: Question About 'ar' Output

[Jon Forrest]

On 4/25/2022 11:15 PM, Jan Beulich wrote:
>
> Member names, when they're longer than what fits in the designated
> member header field, can use an escape mechanism by indexing into the
> special "//" member using the "/<offset>" notation.

Thanks so much for the quick reply.

Can you point me to where this is documented. I haven't been
able to find it.

Is this what's in a "thin" archive?

When you say "indexing into the special "//" member", do
you mean that something like "/47" means 47 bytes after the beginning
of the "//" member?

Also, are the "/" and "//" members actual members or just some
kind of place holders?

Thanks,
Jon

Re: Question About 'ar' Output

[Jan Beulich]

On 26.04.2022 21:17, Jon Forrest wrote:
> On 4/25/2022 11:15 PM, Jan Beulich wrote:
>> Member names, when they're longer than what fits in the designated
>> member header field, can use an escape mechanism by indexing into the
>> special "//" member using the "/<offset>" notation.
> 
> Thanks so much for the quick reply.
> 
> Can you point me to where this is documented. I haven't been
> able to find it.

No, sorry.

> Is this what's in a "thin" archive?
> 
> When you say "indexing into the special "//" member", do
> you mean that something like "/47" means 47 bytes after the beginning
> of the "//" member?

Something along these lines yes. I don't recall off the top of my head
whether some further constant needs adding.

> Also, are the "/" and "//" members actual members or just some
> kind of place holders?

I'd call them "special members".

Jan

Re: Question About 'ar' Output

[Alan Modra]

On Tue, Apr 26, 2022 at 12:17:05PM -0700, Jon Forrest via Binutils wrote:
> Can you point me to where this is documented. I haven't been
> able to find it.

"man 5 ar" or if that doesn't give you anything the good folk of the
freebsd project have documented the format.
https://www.freebsd.org/cgi/man.cgi?query=ar&sektion=5

> Is this what's in a "thin" archive?

No.  A thin archive doesn't store the file contents, instead just
indexing the files on your file system.

-- 
Alan Modra
Australia Development Lab, IBM