This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [MTASCsft PATCH WIP5 00/33] MT-, AS- and AC-Safety docs
- From: "Carlos O'Donell" <carlos at redhat dot com>
- To: Alexandre Oliva <aoliva at redhat dot com>, libc-alpha at sourceware dot org
- Cc: mtk dot manpages at gmail dot com
- Date: Mon, 18 Nov 2013 15:37:11 -0500
- Subject: Re: [MTASCsft PATCH WIP5 00/33] MT-, AS- and AC-Safety docs
- Authentication-results: sourceware.org; auth=none
- References: <20131113081059 dot 3464 dot 51385 dot stgit at frit dot home>
On 11/13/2013 03:11 AM, Alexandre Oliva wrote:
> Here's an updated version of the patch series that documents
> Multi-Thread, Async Signal and Async Cancel safety properties of our
> implementation.
>
> A few files with functions containing lots of dependencies were
> completed since the last round. I modified the commit messages so
> that stgit would indicate the files in the subject of each patch
> email. I think I have clearer definitions for
> uunguard/xguargs/staticbuf, but I'm yet to complete reviewing all uses
> thereof. The new marks are tempsig, stimer, and uplugin.
>
>
> Here's a summary of of the number of @deftypefu?n entries in each file
> in the manual, how many @safety notes there are, how complete the
> safety documentation for that file is, and how many other functions
> I've (so far) noticed as not documented at all in the manual, but that
> should be documented in that file (with a @safety note to go with it)
>
> Filename @def @saf cmplt% undoc
> manual/stdio.texi 126 126 100.00% 0
> manual/arith.texi 102 102 100.00% 0
> manual/string.texi 99 99 100.00% 1
> manual/math.texi 86 86 100.00% 0
> manual/filesys.texi 78 78 100.00% 18
> manual/llio.texi 53 53 100.00% 1
> manual/memory.texi 44 44 100.00% 0
> manual/ctype.texi 36 36 100.00% 0
> manual/process.texi 22 22 100.00% 0
> manual/charset.texi 18 18 100.00% 0
> manual/errno.texi 13 13 100.00% 0
> manual/message.texi 12 12 100.00% 0
> manual/job.texi 11 11 100.00% 0
> manual/pattern.texi 11 11 100.00% 0
> manual/crypt.texi 10 10 100.00% 0
> manual/lang.texi 7 7 100.00% 0
> manual/argp.texi 6 6 100.00% 0
> manual/locale.texi 5 5 100.00% 0
> manual/conf.texi 4 4 100.00% 0
> manual/debug.texi 3 3 100.00% 0
> manual/getopt.texi 3 3 100.00% 0
> ----
> manual/search.texi 15 7 46.67% 0
> manual/signal.texi 27 7 25.93% 0
> manual/resource.texi 31 7 22.58% 0
> manual/terminal.texi 24 5 20.83% 0
> manual/startup.texi 15 3 20.00% 0
> manual/users.texi 66 11 16.67% 0
> manual/time.texi 33 5 15.15% 0
> manual/socket.texi 59 5 8.47% 0
> manual/sysinfo.texi 22 1 4.55% 0
> ----
> manual/platform.texi 8 0 0.00% 0
> manual/setjmp.texi 8 0 0.00% 0
> manual/syslog.texi 5 0 0.00% 0
> manual/pipe.texi 4 0 0.00% 0
> manual/threads.texi 2 0 0.00% 119
> manual/libdl.texi 0 0 0.00% 9
> ----
> Total 1068 800 74.91% 148
These statistics give me goosebumps, I love seeing documenting
get an update like this. We've already found this review userful
internally at Red Hat to answer customer questions, so I'm excited
to see it get checked in.
Overall there have been no objections to these annotations except
those posed by Torvald Riegel and Rich Felker surrounding the
value of the documentation given that MT-safe is a weakened definition
based on the loose language in the POSIX standard meaning data-race
free. The resolution here has been that Torvald and others are going
to work upstream to make sure that we get more precise language into
the standard. I'll be helping with that. It isn't something you need
to worry about unless you are also interested in helping out on that
front in the future. I imagine though that after this review you might
not want to see any documentation work for a while :-)
The only other objection we've had is from Joseph Myers who would
prefer you fix the bugs instead of documenting them. I agree that fixing
bugs is good but that's orthogonal to your purpose with this work and
would impact completing it on time. I'm happy to come back and do bug
fixing after we get this done.
Without any solid objections from anyone, I'm going to do my review
of your patches one-by-one and give you my ACK for these changes.
If you agree, and I agree, and nobody else (Joseph, Torvald, or
Rich) disagrees, we can start checking in these notes for 2.19.
I'm going to explicitly ask Joseph, Torvald, and Rich to comment
again on their approval for this project.
Cheers,
Carlos.