[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

[Git][ocaml-team/approx][master] 18 commits: #1073871 is fixed by pending change



Title: GitLab

Stéphane Glondu pushed to branch master at Debian OCaml Maintainers / approx

Commits:

  • 75b7ae9d
    by Stephane Glondu at 2024-07-02T09:38:27+02:00
    #1073871 is fixed by pending change
    
  • 27ba2525
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Use String.lowercase_ascii and String.capitalize_ascii
    
    Fix deprecation warnings:
    
    ```
    43 |   match String.lowercase str with
                 ^^^^^^^^^^^^^^^^
    Alert deprecated: Stdlib.String.lowercase
    Use String.lowercase_ascii/StringLabels.lowercase_ascii instead.
    ```
    
    ```
    23 |         (String.capitalize field)
                  ^^^^^^^^^^^^^^^^^
    Alert deprecated: Stdlib.String.capitalize
    Use String.capitalize_ascii/StringLabels.capitalize_ascii instead.
    ```
    
  • a4789ee4
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Add a test for config file that only contains blank lines
    
    The config parser fails in this case, as reported in #1031411
    
  • 7e9caf9f
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Remove blank lines when parsing config files
    
    Due to switch from Pcre to Re.Pcre in commit d6dd5bf2, blank lines (ie.
    lines that contain only whitespaces or tabs) now fail the config parser.
    
    More specifically, the issue is caused by the fact that Re.Pcre.split()
    doesn't behave the same as Pcre.split().
    
    Re.Pcre gives:
    
        let whitespace_re = Re.Pcre.regexp "\\s+";;
        Re.Pcre.split ~rex:whitespace_re (" ");;
    
        -- returns --> string list = [""]
    
    While Pcre gives:
    
        Pcre.split (" ");;
    
        -- returns --> string list = []
    
    Reported at https://github.com/ocaml/ocaml-re/issues/288 in case it's a
    bug in the Re library.
    
    In the meantime, just make sure to remove the blank characters from
    lines that contain only that, so that Re.Pcre.split() returns an empty
    list, thus restoring a behavior similar to Pcre.split().
    
    Closes: #1031411
    
  • 61b68fcb
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Update doc/FAQ regarding changing the cache path
    
    Re-instate the original way to change the cache path, from commit
    90c01673. Clarify that there are two ways to change the cache path, and
    they both have their pros and cons.
    
    Also remove the part about approx-gc (this helper doesn't exist
    anymore).
    
  • 0b6fbc0a
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Update doc and config for nowadays debian-security repository
    
    The URL for the debian-security repository changed since the doc and
    default config file were written.
    
    In particular, in /etc/approx/approx.conf, the line:
    
        security http://security.debian.org/debian-security
    
    is not functional (at least for me).
    
    Trying to use approx with the security repo enabled, and via the
    environment variable http_proxy=http://localhost:9999, gives this error:
    
        Connection from ::ffff:127.0.0.1 port 41986
        Request: GET http://security.debian.org/debian-security/dists/bookworm-security/InRelease
          Host: security.debian.org
          Cache-Control: max-age=0
          Accept: text/*
          User-Agent: Debian APT-HTTP/1.3 (2.6.1)
        No remote repository for debian-security  <- THE ERROR!
    
    Now fixed.
    
  • 15e8320d
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Prefer https for documentation links
    
  • 11c699f6
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Add unit tests for release.ml
    
  • 3d7def5a
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Refactor the code for immutable function in release.ml
    
    Refactor so that we have two separate code paths for immutable files:
    
    1) immutable suffixes, and
    2) tarballs
    
    This in preparations for the next commits, so that we can special-case
    some tarballs that are NOT immutable.
    
  • 36c3be91
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Fix: dep11 icon tarballs are NOT immutable
    
    Within the dep11 directory, there are a few tarballs that are not
    immutable: these are the icons-* tarballs.
    
    We can easily single it out based on directory name (dep11) and then a
    filename pattern.
    
    Note that String.starts_with is only available from ocaml 4.13.
    
    Closes: #810255
    
  • 24cb2fff
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Fix: current installer tarballs are NOT immutable
    
    Here's how the installer-*/ directory looks like:
    
    ```
    $ ls -l installer-amd64/
    total 8
    drwxr-xr-x 3 user user 4096 Jun  7  2023 20230607
    drwxr-xr-x 3 user user 4096 Feb  5 14:43 20230607+deb12u5
    lrwxrwxrwx 1 user user   16 Feb 10 16:26 current -> 20230607+deb12u5
    ```
    
    Meaning: the subdirectories are different versions of the installer
    (date-based), so the various tarballs within each of these directories
    are immutables.
    
    EXCEPT, as you can see, the current symlink. Every tarball under the
    current/ are not immutable, and must not be treated as such.
    
    Closes: #757945
    
  • d4ab5672
    by Arnaud Rebillout at 2024-07-02T09:43:55+02:00
    Disable burst limit in systemd socket (rhertzog)
    
    debootstrap makes many requests in a small timeframe (via wget), and the
    approx service gets activated then deactivated for each request.
    
    On a machine where many debootstrap process might run in parallel (say a
    a Jenkins instance running up to 32 jobs at the same time), we hit the
    trigger limit:
    
        approx.socket: Trigger limit hit, refusing further activation.
        approx.socket: Failed with result 'trigger-limit-hit'.
    
    Therefore, with this commit we completely disable the burst limit, as
    stated in the documentation:
    
        $ man systemd.socket
        [...]
            TriggerLimitIntervalSec=, TriggerLimitBurst=
            [...]
            Set either to 0 to disable any form of trigger rate limiting.
    
    Closes: #884713
    
  • ca872c09
    by Stephane Glondu at 2024-07-02T09:43:55+02:00
    Fix fold_test test
    
    It was flaky in concurrent settings. Resetting Config_file before each
    read seems to fix it.
    
  • 6ce6b7a9
    by Stephane Glondu at 2024-07-02T09:44:43+02:00
    Bump version
    
  • de2b7904
    by Stephane Glondu at 2024-07-02T09:46:32+02:00
    Merge tag 'upstream/5.13'
    
    approx version 5.13
    
  • 5b01493f
    by Stephane Glondu at 2024-07-02T09:49:06+02:00
    New upstream release
    
  • e224cbcd
    by Stephane Glondu at 2024-07-02T09:49:06+02:00
    Remove all upstream patches that were applied upstream
    
  • 981d78cb
    by Stephane Glondu at 2024-07-02T09:50:48+02:00
    Prepare upload to unstable
    

30 changed files:

The diff was not included because it is too large.

Reply to: