Re: [PATCH] Add the "-info-os" command to MI

[Kwok Cheung Yeung <kcy@codesourcery.com>]

Hello

Sorry for taking so long to reply!

On 13/10/2011 5:12 PM, Eli Zaretskii wrote:
>>    On @sc{gnu}/Linux, the following values of INFOTYPE are valid:
>>
>> -@anchor{linux info os infotypes}
>> +@anchor{Linux OS information types}
>
> I don't see these lines in today's CVS version of gdb.texinfo.  Does
> your patch depend on some other patch?
>

This patch assumes that the patch from 
http://sourceware.org/ml/gdb-patches/2011-10/msg00368.html has been applied. 
That patch adds many extra types of OS info for Linux only.

> Anyway, why is "info os processes" suddenly treated as a
> Linux-specific command?  "info os" is supposed to be
> platform-agnostic; if it is a Linux-only feature, it will be more
> appropriate to call it "info linux", like "info dos" and "info w32"
> that we already have.
>
>> +The types of information available depend on the target operating
>> +system.  The available types for @sc{gnu}/Linux are listed in
>> +@ref{Linux OS information types}.
>
> As I said above, the referenced section is not Linux-specific, and the
> CVS version of the manual does not have any details there.  So
> something (perhaps in my brain) is missing.
>

In the patch for extra Linux OS data types, I've tweaked the documentation so 
that it says something on the lines of: 'info os INFOTYPE' displays OS 
information of the requested type, types of INFOTYPE available depend on the OS, 
and on Linux, the valid types of INFOTYPE happen to be...	I've moved 'info os 
processes' into that list because I've expanded on the information provided (and 
while the processes INFOTYPE is not specific to Linux, I'm not sure that it is 
universal either?). Anyway, could you take a look at the documentation changes 
in that patch first please?

>> +@smallexample
>> +(gdb)
>> +-info-os
>> +^done,OSDataTable=@{nr_rows="9",nr_cols="2",hdr=[@{width="10",alignment=
>> +"-1",col_name="col0",colhdr="Type"@},@{width="10",alignment="-1",col_nam
>> +e="col1",colhdr="Description"@}],body=[item=@{col0="processes",col1="Lis
>> +ting of all processes"@},item=@{col0="procgroups",col1="Listing of all p
>> +rocess groups"@},item=@{col0="threads",col1="Listing of all threads"@},i
>> +tem=@{col0="files",col1="Listing of all file descriptors"@},item=@{col0=
>> +"sockets",col1="Listing of all internet-domain sockets"@},item=@{col0="s
>> +hm",col1="Listing of all shared-memory regions"@},item=@{col0="semaphore
>> +s",col1="Listing of all semaphores"@},item=@{col0="msg",col1="Listing of
>> + all message queues"@},item=@{col0="modules",col1="Listing of all loaded
>> + kernel modules"@}]@}
>
> The lines here are broken in arbitrary places, which will look
> illegible in the manual.  Please reformat the example so that the
> output could be readable by humans.
>

Well, one could argue that MI wasn't meant to be easily human-readable :-) That 
output is exactly how it would look if the command were issued by a human, and 
it might be confusing if a user issues the command and doesn't get the nicely 
formatted output shown in the manual. I'm happy to make the changes if you think 
that is not an issue though.

Thanks,

Kwok