Patch Package OTP 27.0.1 Released

Patch Package:           OTP 27.0.1
Git Tag:                 OTP-27.0.1
Date:                    2024-07-10
Trouble Report Id:       OTP-19091, OTP-19092, OTP-19094, OTP-19095,
                         OTP-19099, OTP-19100, OTP-19106, OTP-19107,
                         OTP-19108, OTP-19109, OTP-19116, OTP-19118,
                         OTP-19121, OTP-19123, OTP-19131, OTP-19137,
                         OTP-19140, OTP-19142, OTP-19147, OTP-19151,
Seq num:                 ERIERL-1043, ERIERL-1106, GH-8376, GH-8482,
                         GH-8484, GH-8489, GH-8574, GH-8579, GH-8580,
                         GH-8588, GH-8614, PR-8345, PR-8507, PR-8508,
                         PR-8519, PR-8534, PR-8539, PR-8542, PR-8546,
                         PR-8567, PR-8581, PR-8585, PR-8616, PR-8619
System:                  OTP
Release:                 27
Application:             compiler-8.5.1, edoc-1.3.1, erts-15.0.1,
                         kernel-10.0.1, public_key-1.16.1, ssh-5.2.1,
                         ssl-11.2.1, stdlib-6.0.1
Predecessor:             OTP 27.0

Check out the git tag OTP-27.0.1, and build a full OTP system including
documentation. Apply one or more applications from this build as patches to your
installation using the 'otp_patch_apply' tool. For information on install
requirements, see descriptions for each application version below.

# OTP-27.0.1

## Improvements and New Features

- Bump ex_doc version to v0.34.1 and fix copyright in published docs to have
  correct year.

  Own Id: OTP-19095
  Related Id(s): PR-8507

# compiler-8.5.1

The compiler-8.5.1 application can be applied independently of other
applications on a full OTP 27 installation.

## Fixed Bugs and Malfunctions

- One of the compiler's optimization passes would get very slow when compiling
  certain modules. The compiler will now automatically disable that pass for
  input that would trigger the slowdown.

  Own Id: OTP-19131
  Related Id(s): PR-8567

- Fix `+deterministic` to work properly with documentation attributes.

  Own Id: OTP-19142
  Related Id(s): GH-8579, PR-8585

> #### Full runtime dependencies of compiler-8.5.1
> crypto-5.1, erts-13.0, kernel-8.4, stdlib-6.0

# edoc-1.3.1

The edoc-1.3.1 application can be applied independently of other applications on
a full OTP 27 installation.

## Fixed Bugs and Malfunctions

- Fix broken makefile dependency when building HTML documentation.

  Own Id: OTP-19116
  Related Id(s): PR-8534

> #### Full runtime dependencies of edoc-1.3.1
> erts-11.0, inets-5.10, kernel-7.0, stdlib-4.0, syntax_tools-2.0, xmerl-1.3.7

# erts-15.0.1

The erts-15.0.1 application can be applied independently of other applications
on a full OTP 27 installation.

## Fixed Bugs and Malfunctions

- In rare circumstances the JIT could do an unsafe in-place update of a tuple.

  Own Id: OTP-19108
  Related Id(s): PR-8539

- When a port command crashed in the inet driver during gen_tcp:send/2, a
  monitor `'DOWN'` message could be left lingering in the caller's mailbox. This
  has now been fixed.

  Own Id: OTP-19121
  Related Id(s): GH-8484

- `'DOWN'` messages originating from a monitored port, contained the atom
  `process` instead of the atom `port` as the third element when the exit reason
  was not an immediate term.

  Own Id: OTP-19123
  Related Id(s): GH-8484, PR-8546

- Fix so that the options to enable Transparent Huge Page alignment of the
  Erlang VM executable are only applied to the Erlang VM and not other native
  programs such as `erlc` and `dialyzer`. This bug was introduced in Erlang/OTP

  Own Id: OTP-19137
  Related Id(s): GH-8574

- When *no time warp mode* was enabled, a smaller Erlang monotonic time could
  be read than a previously read time, i.e., breaking the monotonic property.
  The runtime system will abort when detecting an issue like this since OTP and OTP 25.0.

  Up until OTP 25 _no time warp mode_ is the default. As of OTP 26 [*multi time
  warp mode*] is the default.

  Own Id: OTP-19147
  Related Id(s): ERIERL-1043, ERIERL-1106, PR-8619

- When calling `trace:function(Session, _, true, meta)` the meta tracer was
  incorrectly set to be the calling process. Now it's set to the session tracer
  as expected.

  Own Id: OTP-19151
  Related Id(s): GH-8614, PR-8616

> #### Full runtime dependencies of erts-15.0.1
> kernel-9.0, sasl-3.3, stdlib-4.1

# kernel-10.0.1

The kernel-10.0.1 application can be applied independently of other applications
on a full OTP 27 installation.

## Improvements and New Features

- Polish the `logger` documentation.

  Own Id: OTP-19118
  Related Id(s): PR-8534

> #### Full runtime dependencies of kernel-10.0.1
> crypto-5.0, erts-15.0, sasl-3.0, stdlib-6.0

# public_key-1.16.1

The public_key-1.16.1 application can be applied independently of other
applications on a full OTP 27 installation.

## Fixed Bugs and Malfunctions

- Fix bug in dnsName constraint check, could cause valid cert to be considered
  bad during path validation.

  Own Id: OTP-19100
  Related Id(s): GH-8482, PR-8508

> #### Full runtime dependencies of public_key-1.16.1
> asn1-3.0, crypto-4.6, erts-6.0, kernel-3.0, stdlib-3.5

# ssh-5.2.1

The ssh-5.2.1 application can be applied independently of other applications on
a full OTP 27 installation.

## Fixed Bugs and Malfunctions

- With this change, race condition between connection closing and automatic
  window adjustment is fixed.

  Own Id: OTP-19109
  Related Id(s): PR-8345

> #### Full runtime dependencies of ssh-5.2.1
> crypto-5.0, erts-14.0, kernel-9.0, public_key-1.6.1, runtime_tools-1.15.1,
> stdlib-5.0, stdlib-6.0

# ssl-11.2.1

The ssl-11.2.1 application can be applied independently of other applications on
a full OTP 27 installation.

## Fixed Bugs and Malfunctions

- Check for TLS-1.3 support should check minimum requirements.

  Own Id: OTP-19094
  Related Id(s): GH-8489

- If both TLS-1.3 and TLS-1.2 is supported and TLS-1.2 negotiated convert
  TLS-1.3 ECDSA schemes to TLS-1.2 hash and signature pairs for increased

  Own Id: OTP-19107
  Related Id(s): GH-8376

- TLS-1.3 negotiation now uses SNI based options correctly instead of ignoring

  Own Id: OTP-19140

## Improvements and New Features

- Make it easier to distinguish between a invalid signature and unsupported

  Own Id: OTP-19091

- Enhance ALERT logs to help understand what causes the alert.

  Own Id: OTP-19092
  Related Id(s): GH-8482

- When the default value for signature_algs is used, default the
  signature_algs_cert to the default value + rsa_pkcs1_sha1 to allow this
  algorithms for certificates but not for the TLS protocol. This is for better
  interoperability. If signature_algs is set explicitly signature_algs_cert must
  also be set explicitly if they should be different.

  Own Id: OTP-19152
  Related Id(s): GH-8588

> #### Full runtime dependencies of ssl-11.2.1
> crypto-5.0, erts-15.0, inets-5.10.7, kernel-9.0, public_key-1.15,
> runtime_tools-1.15.1, stdlib-6.0

# stdlib-6.0.1

The stdlib-6.0.1 application can be applied independently of other applications
on a full OTP 27 installation.

## Fixed Bugs and Malfunctions

- Fix so that missing `-doc({file, File})` files only result in a warning and
  not an error.

  Own Id: OTP-19099
  Related Id(s): PR-8542

- Fixed `json` bugs, json:encode_key_value_list/2 did not generate arrays
  and json:decode/3 did not invoke the user callback for `0`.

  Own Id: OTP-19106
  Related Id(s): GH-8580, PR-8519, PR-8581

> #### Full runtime dependencies of stdlib-6.0.1
> compiler-5.0, crypto-4.5, erts-15.0, kernel-10.0, sasl-3.0

# Thanks to

Frej Drejhammar, Igor Goryachev, Michał Muskała

Hello and many thanks for the release!
File otp_doc_man_27.0.1.tar.gz seems to be missing in GitHub assets.

1 Like

When converting all documentation into Markdown, we did not implement a man page backend, reasoning that having man pages for Erlang modules wasn’t a priority. Therefore there is no man page tarball.

But we have realized that we should have man pages for all installed Unix executables, so we will have to fix that, but it won’t happen until sometime after the vacation period in Sweden, that is after August…


Oh wow, have I missed something in all the ex_doc excitement? Does that mean there will be no man page module docs from R27? Or that there will be but they aren’t there yet?

I realise I’m likely an outlier here, but for me personally man pages for modules are at least as high a priority as web pages. I hope they’re not disappearing.

(Edit: ah, maybe implementing the man backend for the CLI tools will mean this happens anyway?)

1 Like

With the documentation chunks in .beam files and the Erlang shell improvements we have function and module documentation in the Erlang shell. This was a much requested feature and I guess that for many users it is a good enough or superior replacement for the Unix man pages, just not in the same shell. It also allows documentation to be presented by all kinds of development environments…

IIRC (the chief responsible are on vacation), we intend to create man pages for the Unix executables, also for OTP 27, in time.

Thanks Raimo. I read about that and it sounds useful, but I had no idea implementing the new doc components would mean actually removing the man pages :scream:

I just tried out the shell version in 27, and as far as I can tell the docs available are either m(sets), h(sets,add_element) or ht(sets,set[,1]). I could be missing things but it looks like you have to do 3 separate things in order to get info about a module and a specific function with types, or 2 even just to look up a module’s functions and then see the doc for one of them, rather than just e.g. man sets -> space -> space or man sets -> /add_el in the standard man/less pager.

Also h(queue,in) only shows in(Item, Q1) for the signature, whereas the function section in the man page shows in(Item, Q1 :: queue(Item)) -> Q2 :: queue(Item), so once you’ve found it, you don’t then need to say “OK, now show me ht(queue) … ok, ht(queue,queue,1)”. Maybe queue is a bad example because it’s got a limited set of quite obvious types, but I’m sure you get the point, it’s quite unwieldy by comparison.

However this isn’t the main issue, which is that I don’t think this comes close to replacing the ability to quickly pull up e.g. man maps in a separate unix shell/tmux pane to check something while coding, and leave it up for reference while needed. Not only that, pulling up a page of module doc in an erlang shell you’re working in would affect the shell flow pretty significantly. You could of course run a separate erlang shell in another pane and use that purely for documentation, but it doesn’t show the full module doc as easily, and it’s fragmented into these different h() functions, so it seems like a bit of a strange thing to have to use a full erlang shell for, rather than just quickly checking the man page.

More importantly, man provides a well-known and fast searchable pager. So you can just say e.g. man sets and scan quickly through the functions (with type info inline), search for a specific term, page up and down as needed - basically just do all the unix doc stuff that man pages have always provided.

I find this invaluable as often I’ll be writing a module which wraps a specific piece of functionality, e.g. providing a bit of extra functionality around a queue, and having the man page for queue up in a shell pane while I work on it lets me get super quick access to the info I need without having to keep all the function signatures in long-term memory so as to be able to call up the doc for the right function. There is the autocomplete on modules and functions, but again, this is really fragmenting what you have to do to get to the info you need.

As with all big changes I guess I’ll get used to it over time but these do seem like downgrades in how to get module/function/type info quickly and efficiently.

Was leaving in the man doc backend deemed such a large maintenance burden? Is it something that a sufficiently motivated person could add back in? Where should I look to try and understand what would be involved?


This will have to wait until the chief responsible are back from the vacation. I think the problem lies in Markdown to Man conversion and thereby the choice of Markdown parser.

I use a workflow much as you describe, but I open a browser tab and view the HTML module documentation. The links between functions and types is something I find very useful.

Thanks @raimo that makes sense. I’ll look forward to when the other folks are back and hopefully find out a bit more. Appreciate your response.