4.17 release plan

[alejandro-colomar]

Cc: @hallyn , @ikerexxe , @thesamesam , @zeha , @jubalh , @stoeckmann

Hi!

We will soon release 4.17.0, around Christmas/Solstice.
The first RC (release candidate) will be out around St. Nick's day.

Here's a set of PRs, in the order of priority (IMO):

---

Pending:

• lib/sgetgrent.c: sgetgrent(): Fix use-after-free bug #1145
  Fixes a use-after-free bug.

• readlinknul(): Minor tweaks #1107
  Trivial stuff; fixes a return type.

• lib/, src/: Use streq() instead of its pattern #1142
  Trivial stuff. Semantic patch provided for easier verification.

---

Queued for after the release of 4.17.0:

• lib/, src/: Use streq() instead of its pattern #1136
  Trivial stuff. A future patch will depend on this one.

• rmdir_leading() improvements #1135
  It is interesting beacuse it adds a string-handling API that I'll use soon in a few patches.

• Use cc(1)'s __has_include() instead of build-system checks #1118
  Trivial stuff. Remove autotools-based header checks, and replace them by a common cc(1) feature.

• configure.ac, lib/, src/: Presume working shadow group support in libc #1111
  Trivial stuff. Remove some build-system check for an old libc bug that was present in the times of Red Hat 3.

• configure.ac, lib/, src/: Assume setgroups(2) and initgroups(3) exist #1125
  Trivial stuff. Remove obsolete check.

• lib/: Some improvements to sizes of array parameters #1134
  Array-bounds safety improvements. Trivial stuff, but triggers CodeQL false positives, which I've dismissed.

• lib/atoi/a2i/: Simplify these macros by calling a2i() #1137
  Trivial stuff. Simplifying liba2i macros.

• lib/fs/mkstemp/, src/: Move fmkomstemp() to separate files under lib/fs/mkstemp/, and split into mkomstemp() #1139
  Move API to lib/fs/ and split into two APIs.

• lib/, src/: Simplify code that calls [sg]etgroups(3) #1123
  Contains a lot of patches simplifying mostly 2 functions. Each patch is more or less trivial. But it has too many patches.

---

Merged already:

• lib/idmapping.c: Fix get_map_ranges regression #1117
  Fix input validation. Prevents introducing a regression.

• Replace most strtok(3) calls by strsep(3) #1093
  This one needs very careful review. If you have the slightest doubts, please skip it. Still, it's interesting because I have many patches that depend on this one in my local tree.

• Simplify #1109
  Contains a lot of patches simplifying mostly a single function. Each patch is more-or-less trivial, but there are many. I have some other patches after this one, so it'd be interesting to have it done before the release.

• contrib/, lib/, src/: Use streq() instead of its pattern #1128
  Trivial stuff. Semantic patch provided for easier verification.

• lib/gshadow_.h: __STDC__ is always 1 #1110
  Trivial stuff. Remove fall-back code for pre-ANSI C, which is now dead.

• src/login_nopam.c: resolve_hostname(): Use NI_MAXHOST instead of MAXHOSTNAMELEN with getnameinfo(3) #1101
  Trivial stuff. Just some pedantic POSIX stuff.

• csrand(): Use POSIX read(2) instead of the heavier stdio #1119
  Some simplification of csrand(). Just review if everything else is done, but we can live without it.

[ikerexxe]

We will soon release 4.17.0, around Christmas/Solstice. The first RC (release candidate) will be out around St. Nick's day.

Is St. Nick's December 6? I am not used to using the calendar of saints.

Here's a set of PRs, in the order of priority (IMO):

It's good to have this list, as there are lots of PRs to take care of

[alejandro-colomar]

We will soon release 4.17.0, around Christmas/Solstice. The first RC (release candidate) will be out around St. Nick's day.

Is St. Nick's December 6? I am not used to using the calendar of saints.

Yes, it is. :)

[zeha]

We will soon release 4.17.0, around Christmas/Solstice.

Thanks for the heads up!

[alejandro-colomar]

There's a new PR from a contributor.

Since all of my PRs are just readability refactorings, please queue them after anything not mine, of course.

[alejandro-colomar]

The release candidate is out. Please try it out! :)

I've converted all of my PRs to draft until we release 4.17.0.

[alejandro-colomar]

I've reopened my PRs for after 4.17.0 already, since the release is imminent.

[hallyn]

So how are we feeling? I didn't do a pre-release last week... Should I do a prerelease
today before final release Tuesday or Wednesday, or just wait for the final release?

[alejandro-colomar]

So how are we feeling? I didn't do a pre-release last week... Should I do a prerelease today before final release Tuesday or Wednesday, or just wait for the final release?

We haven't added much stuff. I would be happy with a final release already.

The changes are:

$ git log --graph --oneline master 4.17.0-rc1 -8
* 8821d3ff (shadow/master, master) lib/fs/readlink/: readlinknul(): Fix return type
* b9d00b64 lib/fs/readlink/readlinknul.h: readlinknul(): Silence warning
* 205c23bf Added option -a for listing active users only, optimized using if aflg,return
* c3930556 lib/, src/: Use !streq() instead of its pattern
* 7182d640 lib/, src/: Use streq() instead of its pattern
* 8424d7c4 lib/sgetgrent.c: sgetgrent(): Fix use-after-free bug
* b75ea298 Release 4.17.0-rc1
| * 059f3bb7 (tag: 4.17.0-rc1) Release 4.17.0-rc1
|/  

[ikerexxe]

I'm fine with doing a final release directly. 4.17.0-rc1 has been available in fedora rawhide for two weeks already and nobody complained, so I don't expect many trouble.

[alejandro-colomar]

Cc: @hallyn , @ikerexxe , @thesamesam , @zeha , @jubalh , @stoeckmann

Hi!

We will soon release 4.17.0, around Christmas/Solstice. The first RC (release candidate) will be out around St. Nick's day.

Here's a set of PRs, in the order of priority (IMO):

[...]

Queued for after the release of 4.17.0:

* [lib/, src/: Use streq() instead of its pattern #1136](https://github.com/shadow-maint/shadow/pull/1136)
  Trivial stuff.  A future patch will depend on this one.

* [rmdir_leading() improvements #1135](https://github.com/shadow-maint/shadow/pull/1135)
  It is interesting beacuse it adds a string-handling API that I'll use soon in a few patches.

* [Use cc(1)'s __has_include() instead of build-system checks #1118](https://github.com/shadow-maint/shadow/pull/1118)
  Trivial stuff.  Remove autotools-based header checks, and replace them by a common cc(1) feature.

* [configure.ac, lib/, src/: Presume working shadow group support in libc #1111](https://github.com/shadow-maint/shadow/pull/1111)
  Trivial stuff.  Remove some build-system check for an old libc bug that was present in the times of Red Hat 3.

* [configure.ac, lib/, src/: Assume setgroups(2) and initgroups(3) exist #1125](https://github.com/shadow-maint/shadow/pull/1125)
  Trivial stuff.  Remove obsolete check.

* [lib/: Some improvements to sizes of array parameters #1134](https://github.com/shadow-maint/shadow/pull/1134)
  Array-bounds safety improvements.  Trivial stuff, but triggers CodeQL false positives, which I've dismissed.

* [lib/atoi/a2i/: Simplify these macros by calling a2i() #1137](https://github.com/shadow-maint/shadow/pull/1137)
  Trivial stuff.  Simplifying liba2i macros.

* [lib/fs/mkstemp/, src/: Move fmkomstemp() to separate files under lib/fs/mkstemp/, and split into mkomstemp() #1139](https://github.com/shadow-maint/shadow/pull/1139)
  Move API to lib/fs/ and split into two APIs.

* [lib/, src/: Simplify code that calls [sg]etgroups(3) #1123](https://github.com/shadow-maint/shadow/pull/1123)
  Contains a lot of patches simplifying mostly 2 functions.  Each patch is more or less trivial.  But it has too many patches.

@hallyn , @ikerexxe , this is still an accurate list of importance of the PRs still open after the release, for when you have some time for reviewing stuff. Especially, the first two PRs are the most important. Merry Christmas! :)

[jubalh]

I also packaged RC 1 on 6th December and it seemed fine.
Today I updated the openSUSE Tumbleweed package to 4.17.0 :)

[hallyn]

@jubalh @zeha @ikerexxe Note the 4.17.1 release today, with 'su -l' regression fix.

I'd send another announcement, but I'm really not sure anyone received the last one :)

[alejandro-colomar]

@jubalh @zeha @ikerexxe Note the 4.17.1 release today, with 'su -l' regression fix.

Cc: @thesamesam (but he's still on 4.14, though).

I'd send another announcement, but I'm really not sure anyone received the last one :)

It's Christmas. Maybe they are vacation-ish. :)

[zeha]

I didn't get an email, maybe I'm not subscribed to the right place.

The github pings are useful. Thanks!

[jubalh]

@jubalh @zeha @ikerexxe Note the 4.17.1 release today, with 'su -l' regression fix.

Already packaged for openSUSE 1 min after the release happened ;)

[alejandro-colomar]

4.17.2 coming soon.