OpenBSD Journal

In -current, default write format for tar(1) changed to "pax"

Contributed by rueda on from the pax-in-our-time dept.

A series of commits by Jeremie Courreges-Anglas (jca@) has modified tar(1) such that its default write format (for archives) is that of pax(1). The message with the final commit captures the gist of the change:

CVSROOT:	/cvs
Module name:	src
Changes by:	jca@cvs.openbsd.org	2024/04/16 17:09:35

Modified files:
	bin/pax        : options.c tar.1 

Log message:
Switch tar(1) write default format to 'pax'

Lets us store longer file names, link names, finer grained timestamps,
larger archive member files, etc; at the expense of larger uncompressed
archives and less widespread support across the ecosystem.  If you're
unhappy with the new defaults, you can use -F ustar.  Or you can help
fix bugs / find a better middle ground.

Prodding from various including job@ and deraadt@
ok sthen@ caspar@ millert@

Reply

Comments

  1. By John McCue (jmcunx) jmcq66@comcast.net on

    Very nice. I ran into the long file issue on one dir structure I was archiving due to a long fie name. I had to use gtar in for that Dir.

    Thanks

    Reply

    Comments

    1. By Sebastian Rother (2001:9e8:f94:8e00:609b:a2bf:81c1:4d21) gsr@privat.phoenix-it.org on

      I did not yet checked the most recent CURRENT but:

      In the Past the user who created a TAR was always VISIBLE so it was always able to determine an existing User-Account.

      I reported this ~2 Decades ago.

      I have other Issues with OpenBSD currently (drivers, iwn and co) and would like to support the Developers by reporting Issues if they would at least take a look.

      I think OpenBSD would benefit by stopping to implement ANYTHING NEW for ONE RELEASE (maybe 2) and fix Issues first.

      Sure, it is not what most Developers want but: RAL Driver is BS, IWN makes Issues (I got a "connection" wich does not work and I have to ifconfig down/up+dhclient the Interface on a "SUPPORTED" WLAN-Device).

      About me: I find Bugs.... like in the Installer! I can render you ANY OpenBSD-Installation USELESS if you do use Encryption! Logic-Bugs...

      So I think focusing on BUG FIXING might should be a Priority! Also the Foundation should REACH OUT to Companies and ask actively for DOCUMENTATION for HARDWARE and being allowed to PUBLISH them so that ANYONE could improve Drivers related on the Documentation of the Vendor (if possible).

      So the Foundation could host a Library for such Documentation (wich they are allowed to collect and publish so they are openly avaiable).

      We even could "Team Up" with the LINUX Guys related to this.

      Reply

      Comments

      1. By Sebastian Rother (2001:9e8:f94:8e00:609b:a2bf:81c1:4d21) gsr@privat.phoenix-it.org on

        I comment to myself: THE USB STACK is BS TOO!!!!
        For like 3 DECADES NOW!

        There USB-FUZZERS avaiable. Some Developers might take a look!
        NTFS (IN KERNEL) is BROCKEN and can COMPROMISE (my Oppinion..) the OS by using it.
        I send multiple Reports too.

        So attaching a NTFS-formated USB-STICK might be dangerous!
        Why not dropping it all! Alternative: USE FUSE but FUSE needs to be SUPPORTED.... another Issue. Am I a FAN of FUSE? No: But at least then you don't get crashed or compromised....

        I am for DROPPING AS MUCH AS POSSIBLE and going "all in" for FUSE since OpenBSd has NOBODY who maintains FAT, NTFS and other Filesystems.

        And yes I reported MULTIPLE CRASHES related to NTFS (in Kernel) by JUST USING CLAMAV to scan a Device....

        I did not even needed a Fuzzer to Crash OpenBSD. All I needed was a NTFS-HDD and ClamAV. Is this the STABILITY Theo talks about?

        Reply

        Comments

        1. By Anonymous Coward (185.42.170.203) on

          I also wonder what is the point of the read-only NTFS driver in OpenBSD, when there is a FUSE file system for that. People who want to recover things from NTFS partitions would be better off doing it using a different OS like Windows, or even Linux which can deal with NTFS better.

          OpenBSD is great for certain network devices, but for my usage (and likely many others too) it lacks in a few things:
          - No file system integrity protection measures such as journaling, and now no softdep. And fsck doesn't count.
          - No TRIM support for SSDs (which is a necessity, contrary to copium addicts).
          - No proper NVMe support (it goes through the SATA translation layer which visibly hurts performance)
          - Poor virtualization support (for now, anyway)
          - No support for TPMs and OPAL.
          - Some rough edges, eg. the whole system hangs if a mounted USB is unplugged, rather than just cleaning up and unmounting automatically and just crashing programs that may still be using the mount point or files within.
          - A chrony port would be nice for very accurate timekeeping and sync.

          Of course, Linux has its own fair share of bugs and issues which makes it difficult to justify using on a router or such appliance, where OpenBSD shines.

          As far as OpenBSD teaming up with the Linux guys, I often find the Linux guys are even less welcoming than the OpenBSD guys, unless you're a gender-confused freak from a major corporation whose sole purpose in life is to make people miserable and smoke hashish. If the problems listed above were to be fixed I wouldn't hesitate to move my desktops over to OpenBSD.

          Reply

          Comments

          1. By Anonymous Coward (2620:7:6001::187) on

            To add to the FS issue also, Windows has a nice pluggable file system driver interface, to allow other file systems to be supported in a modular fashion, while the others prefer stuffing all their file systems in a monolithic way in the kernel. Want to support multiple file systems properly in kernel space, follow the Windows way.

            Reply

      2. By Anonymous Coward (49.12.42.182) on

        Sebastian, getting due credit is nice. But don't let it kill you when you don't get credit. Sure it's happened to everyone!

        I remember decades ago (musta been 25 years or more) when I was on #freebsd on efnet asking about the "garble" that came with the padding of ICMP packets. It was told to me that "this is just kernel internal data and noone can make use of that" or something along those lines. Well OpenBSD had it's eyes and ears open in those channels I gather because when it came out with zero padding packets that were under the 64 byte size in ethernet, I was excited that my question and concern was finally addressed.

        I was a newby then and let it go, but it helped me a lot to get confirmation that this was not kosher.

        As a hint you may want to tone it down a little and take it easy about this issue. For you it may have more relevance than to others. In the big picture things will get fixed... eventually, and for credit of someone else. :-)

        -pjp

        Reply

        Comments

        1. By anon (anonymouse) on

          There are lots of bug reports for various things and limited developer resources to deal with them. Those from people who give the information needed to identify the problem and presented in a nice easy-to-read way tend to get looked at first. Those from reporters who are aggressive, often sending incoherent messages, and seeming quite entitled, are more likely to get ignored or looked at later, it's just generally not a pleasant experience working with them.

          Reply

      3. By System Operator (46.10.166.16) on

        Obviously, your rate of reporting to resolution is very low. To improve on this:
        1. report accurately with technically correct details
        2. behaviour of the user are not system faults
        3. find a local support group
        4. find another system
        5. stop whining

        Reply

  2. By Sebastian Rother (2001:9e8:f94:8e00:609b:a2bf:81c1:4d21) gsr@privat.phoenix-it.org on

    I will Theo NEVER forget that he never ACCAPTED my BUG as the 3rd SECURITY HOLE in a DEFAULT INSTALLATION.

    I once Reported a Bug in PF, in "CURRENT", close to become STABLE.
    THIS IS WHY I WAS NEVER NAMED....

    Henning (back then) confirmed it, others too...
    Just Because of Theoes Ego he dishonored me.

    OpenBSD had 3 SECURITY HOLES in a DEFAULT INSTALLATION...
    Because: I reported the ipv4/ipv6 Header (ICMP) Issue back then as OpenBSD was on its way to ENABLE PF by DEFAULT.

    Theo said: IT WAS NEVER SHIPPED THIS WAY.
    Strange.. and how did "CURRENT" back then behaved? NOBODY ENABLED IT?

    I just like to point this out to gain some Credit for my other Posts.
    THERE ISSUES in the Code, and well: We all make Issues!
    But there some Issues I reported DECADES AGO and wich where IGNORED by leading OpenBSD Developers because they think I am a Troll.

    I am not...

    Reply

    Comments

    1. By Anonymous Coward (2001:9e8:f94:8e00:609b:a2bf:81c1:4d21) on

      To clarify: i named ICMP but it was the IP HEADERS. I send IPv6 Headers with IPv4 Packets: Kernel CRASH... later they investigated the ICMP-STACK again wich I did not reported since I was simply dropped and "silenced".

      You can Crash OpenBSD using ClamAV or using USB....
      Seriously Theo, and I hope somebody sends you this you South African, YOU ARE WHERE YOU STARTED! Whole Code needs a REVIEW! CONGRATIULATIONS THEO! This is what you once wanted to prevent!

      IWN Crashes your DEVICE? What did you told me: "POOR ME" or so? At least you commented on the DMESG I send ya. Is an INTEL N100-Platform so "seldom" and could lead to CRASHES that you PERSONALY ignore my Reports?

      Well, you life in Canada, the Land of WOKE.... this shows.
      As you started, you took Bug-Reports seriously but this was like 3 Decades ago. I still will remmeber you; being an inspiring Guy to me. You found logic Bugs in the INTEL CORE Architecture (HARDWARE! wich remains uppon today and some are not fixed anyway). I still remmeber this.

      Still... others might find Bugs too. Why do you ignore the simply truth that OpenBSD needs a COMPLETE REVIEW! With modern Technologies like FUZZERS and co! Sure: SOME Developers use some but what I named: USB-Stack, WLAN-Stack (RAL Drivers and some others, IWN for example...)...

      And NO: I wont become a DEVELOPER ("Send us PATCHES!"): I need to support somebody myself and have no time for this (this is very private). but I can find Bugs and report them or lend/buy Developers the Gead I use to replicate it..

      Reply

      Comments

      1. By bodie (2a00:ca8:a15:9024:ca5a:cfff:feb0:2154) on

        I would say that even Theo changed during a years and what you are describing is perfectly possible back then, but less today as whole *BSD community moved towards more cooperation when and where possible ( some past tensions may be not possible to remove of course )

        - ClamAV ... you would not believe how many crashes and problems I saw in production with so called professional AV solutions on so called great Linux systems. ClamAV is one of the less problematic, but is still from Cisco. In the end it is AV. That some is not currently crashing on OS of your choice does not mean that it runs properly internally ;-)

        - USB ... here I feel devs are aware and moving it forward

        - WLAN ... well other BSDs are in a need for improvement here as well, it is connected to man power and availability of documentation. Sure you can allow vendors directly in kernel like in case of Linux, but then when you look around what they make there you want to throw up :-D

        - Filesystems ... never had issues with OpenBSD even after crashing or loosing power on production systems as well. Yeah, worked a lot in production with ZFS, now with btrfs, of course tested Hammer and such. But is still usable and proven even that ffs. And actually a lot of interesting work is happening with removal of softdep which may help bring more people in to this area or even import some new and/or existing filesystem from elsewhere

        It's possible to continue on and on. It was always clear that OpenBSD was primarily for its devs and if you find it suitable good for you and if not then shut up and hack or leave. May sound harsh, but years show that it really works. Always have it on private desktop/laptop or some devices plus where I can't have directly then at least in virtualization as the source of some documentation and reference point (or sanity point often these days :-D)

        Just yesterday tested NetBSD 10 on real HW. They really did a lot of work, but they have systems which needs some love too like eg. that old Bluetooth stack which was already removed from OpenBSD (and yet you still can use BT mouses, keyboards,....). Working with services is less "clean" compared to OpenBSD, especially those from packages, but no show stopper. Sure there is a bug in dhcp client timing out attempt to display lease, but that is reported bug inside dhcpd itself so external component. They have great things like rumpkernel, still have Kerberos or PAM for easier integration in to Linux/Windows-centric environment, still have Linux compatibility layer (which needs a lot of work too). Their lack of man power is of course worse compared to OpenBSD and yet they still continue to do such great things

        DragonflyBSD ... well latest snapshot do not see my WiFi at all (OpenBSD/NetBSD just fine, but this was always the case during the past years). Starting X freezes laptop completely so no keyboard/mouse inside, but you can't even switch back to console. At least ACPI button works so that you can send shutdown signal :-D

        FreeBSD ... never got its taste for multiple reasons. Mainly because I prefer simplicity and that it reminds me Linux too much already. I may be completely wrong of course and nothing against its devs. I am just used to something else and that is probably bigger reason. In the end if Netlix can use it for sharing worlwide and at least in the past WhatsApp can used that for their infrastructure then they do something right as well :-)

        Reply

    2. By Anonymous Coward (113.190.176.207) on

      You may leave OpenBSD, the only operating system that you are not the god if you dont code.
      Why pointing out its "issues" to make it become much more competitive to linuxx? you are dumber than a kid. and animals dont even do things like you are doing now.

      Reply

  3. By Anonymous Coward (49.12.42.182) on

    So does this mean we should by default name our tarballs (paxballs)
    --> filesarchive.pax.gz? in order to avoid confusion?

    Thanks!
    -pjp

    Reply

    Comments

    2. By Anonymous Coward (138.199.48.186) on

      do you name your current tarballs (ustarballs) filesarchive.ustar.gz? in order to avoid confusion?

      Reply

      Comments

      1. By Anonymous Coward (2a01:4f8:c012:fec9::1) on

        Is this supposed to be ridicule(tm)? *plonk*

        Reply

    3. By System Operator (46.10.166.16) on

      Avoid being called out utterly stupid and illiterate more than once a month, or hourly for the incapacitated..

      The compressed archive format file extension for the OpenBSD release sets changed, from tar.gz to tgz with the
      turn of the Millenium for OpenBSD 2.7 relesed on 2000-June-15, see https://www.openbsd.org/27.html and compare

      listing https://ftp.eu.openbsd.org/pub/OpenBSD/2.6/i386/ with https://ftp.eu.openbsd.org/pub/OpenBSD/2.7/i386/
      Yes, the "june" bug has been biting you for such long eternity, you should have developed a chemical reaction.

      One more recent version has been summarised by the ever changing and never stable continuosly imperfect online
      Frequently Asked Questions 4. Installation Guide, File Sets https://www.openbsd.org/faq/faq4.html#FilesNeeded

      You also as an amateur computer consumer enthusiast understand the relation of requirement, dependence and use
      so you know that whatever it is used, the file extension is in the preparation scripts and the installer ones.

      Thus naturally an eager to learn mind is inclined to look into the nethers of ever changing repositories of it
      likes the installer ALLSETS are belong to https://cvsweb.openbsd.org/src/distrib/miniroot/install.sub?annotate

      For the more "stupider" of you, check around .tgz as the opening paragraph has been postulating happened ~Y2K.
      After 25 years of the New Age of computing post legacy compute, and further 25 years of BSD before, what is it
      going to be called in 25 years in 2050, this mythical archive magic byte undefined file extension of, easy .pz

      https://en.wikipedia.org/wiki/Tar_(computing)?useskin=vector#Suffixes_for_compressed_files
      https://en.wikipedia.org/wiki/Tar_(computing)?useskin=vector#POSIX.1-2001/pax
      https://en.wikipedia.org/wiki/Pax_(command)?useskin=vector#Format_support

      For a glimmer of hope, Linux will still be stuggling for some adoption, and you might still be having Internet
      & that's about the best of what you can glimpse of the future without paying the Oracle some Big Tech Dollars.
      By the way, there is no US or Dollar in 2050, and everything you have learned from the Internet is BS in 2030.

      Reply

Latest Articles

Credits

Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original deadly.org with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]