• postfix-3.11.4,1 does not work with lmdb-1.0.0,1, requires 0.93.5,1

    From Jordan M@freebsdlists@montesse.ca to muc.lists.freebsd.ports on Fri Jul 3 16:47:36 2026
    From Newsgroup: muc.lists.freebsd.ports

    Following up from an issue raised over with Arch Linux[1] about postfix,
    knot, bogofilter, et al not working with lmdb 1.0.0, I tested on FreeBSD
    and confirmed lmdb-1.0.0 silently breaks consumers that work with 0.93.5
    in a way not fixed by rebuilding after a library version bump.

    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a
    separate origin?


    Building ports as of commit 368aa8c31354 (earlier today) on amd64
    15.1p1, postfix builds successfully, but postmap cannot create lmdb
    databases:

    # pkg info -d postfix | grep lmdb
    lmdb-1.0.0,1
    lmdb-1.0.0,1 (liblmdb.so.1)
    # postmap lmdb:/tmp/testdb
    postmap: fatal: error updating lmdb:/tmp/testdb: MDB_BAD_TXN:
    Transaction must abort, has a child, or is invalid
    #

    MDB_BAD_TXN is the same error as reported on Arch. There are apparently
    other incompatible changes as well, but this is the easy one to spot
    with postfix.


    If I build from before the update, postmap works again:

    # git log -1 --oneline databases/lmdb/Makefile
    1d1d6afb80a9 databases/lmdb: upgrade to 1.0.0
    # git checkout 1d1d6afb80a9^1
    HEAD is now at 3c288518d2cf devel/bazel9: upgrade to 9.1.1.
    [construction noises]
    # pkg info -d postfix | grep lmdb
    lmdb-0.9.35,1
    lmdb-0.9.35,1 (liblmdb.so.0)
    # postmap lmdb:/tmp/testdb
    #


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From =?UTF-8?Q?Einar_Bjarni_Halld=C3=B3rsson?=@einar@isnic.is to muc.lists.freebsd.ports on Fri Jul 3 23:59:27 2026
    From Newsgroup: muc.lists.freebsd.ports

    On 7/3/26 11:47 PM, Jordan M wrote:
    Following up from an issue raised over with Arch Linux[1] about postfix, knot, bogofilter, et al not working with lmdb 1.0.0, I tested on FreeBSD
    and confirmed lmdb-1.0.0 silently breaks consumers that work with 0.93.5
    in a way not fixed by rebuilding after a library version bump.

    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a
    separate origin?


    I agree with this. Upgrading lmdb broke both knot3 and bind920 for me.
    I've fixed it by downgrading lmdb in a ports-overlay tree that poudriere
    uses.

    .einar


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From A FreeBSD User@freebsd@walstatt-de.de to muc.lists.freebsd.ports on Sat Jul 4 10:27:02 2026
    From Newsgroup: muc.lists.freebsd.ports

    --Sig_/vQrTnn6U8VLm2G5V=STiBG6
    Content-Type: text/plain; charset=UTF-8
    Content-Transfer-Encoding: quoted-printable

    Am Tage des Herren Fri, 3 Jul 2026 23:59:27 +0000
    Einar Bjarni Halld=C3=B3rsson <einar@isnic.is> schrieb:

    On 7/3/26 11:47 PM, Jordan M wrote:
    Following up from an issue raised over with Arch Linux[1] about postfix=
    ,=20
    knot, bogofilter, et al not working with lmdb 1.0.0, I tested on FreeBS=
    D=20
    and confirmed lmdb-1.0.0 silently breaks consumers that work with 0.93.=
    5=20
    in a way not fixed by rebuilding after a library version bump.
    =20
    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a=20 separate origin?
    =20
    =20
    I agree with this. Upgrading lmdb broke both knot3 and bind920 for me.=20 I've fixed it by downgrading lmdb in a ports-overlay tree that poudriere=
    =20
    uses.
    =20
    .einar
    =20

    ... for net/bind920: I use the "make" method building ports and I had to re= build recursiveley
    all depending ports for net/bind920 via "portmaster -f" - not identifying e= xactly which port
    seems to have trouble with new lmdb - but this proceeding helped fixing the=
    issue. In
    poudriere, the way is similarily radical.

    --=20

    A FreeBSD user

    --Sig_/vQrTnn6U8VLm2G5V=STiBG6
    Content-Type: application/pgp-signature
    Content-Description: OpenPGP digital signature

    -----BEGIN PGP SIGNATURE-----

    iJEEARYKADkWIQRQheDybVktG5eW/1Kxzvs8OqokrwUCakjD8RsUgAAAAAAEAA5t YW51MiwyLjUrMS4xMiwyLDMACgkQsc77PDqqJK+WTQD/eTvzfHKuNm9IH6s65jyh MzunO6G6VFirNn71ml21qu0A/Rc7vpo+QxZtM1jTpBVxw1wRkIOy4NGiCZ36IMmj
    fIAO
    =uUlA
    -----END PGP SIGNATURE-----

    --Sig_/vQrTnn6U8VLm2G5V=STiBG6--


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Michael Grimm@trashcan@ellael.org to muc.lists.freebsd.ports on Sat Jul 4 11:09:15 2026
    From Newsgroup: muc.lists.freebsd.ports

    Einar Bjarni Halld||rsson <einar@isnic.is> wrote:
    On 7/3/26 11:47 PM, Jordan M wrote:
    Following up from an issue raised over with Arch Linux[1] about postfix, knot, bogofilter, et al not working with lmdb 1.0.0, I tested on FreeBSD and confirmed lmdb-1.0.0 silently breaks consumers that work with 0.93.5 in a way not fixed by rebuilding after a library version bump.
    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a separate origin?


    I agree with this. Upgrading lmdb broke both knot3 and bind920 for me. I've fixed it by downgrading lmdb in a ports-overlay tree that poudriere uses.
    Thanks to Jordan for the heads up and thanks for your solution with a port overlay.
    I always wanted to try overlays, now it was time to implement an overlay ports tree. Not only for databases/lmdb and all my other modifications/additions as well. Nice feature.
    Regards,
    Michael
    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Michael Grimm@trashcan@ellael.org to muc.lists.freebsd.ports on Sat Jul 4 11:40:32 2026
    From Newsgroup: muc.lists.freebsd.ports

    Jordan M <freebsdlists@montesse.ca> wrote:
    Following up from an issue raised over with Arch Linux[1] about postfix, knot, bogofilter, et al not working with lmdb 1.0.0, I tested on FreeBSD and confirmed lmdb-1.0.0 silently breaks consumers that work with 0.93.5 in a way not fixed by rebuilding after a library version bump.
    Now, I found in UPDATING (should have looked earlier):
    <snip>
    20260702:
    AFFECTS: users of databases/lmdb
    AUTHOR: delphij@FreeBSD.org
    LMDB 1.0 introduced an incompatible on-disk file format change.
    Versions 0.9.x and 1.0.x databases are mutually incompatible.
    Before upgrading, export all existing databases using the old v0.9
    mdb_dump utility, then import them with the new v1.0 mdb_load after
    upgrading. There is no support for opening v0.9 database files
    directly with LMDB 1.0.
    Example migration procedure:
    # mdb_dump -a /path/to/db > /tmp/mydb.dump
    # pkg upgrade databases/lmdb
    # mdb_load -f /tmp/mydb.dump /path/to/newdb
    <snip>
    That would be a whole lotta work for my numerous knot3 and postfix-current lmdb databases at some servers ...
    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a separate origin?
    +1
    Regards,
    Michael
    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Gert Doering@gert@greenie.muc.de to muc.lists.freebsd.ports on Sat Jul 4 11:45:42 2026
    From Newsgroup: muc.lists.freebsd.ports

    Hi,

    On Sat, Jul 04, 2026 at 11:40:32AM +0200, Michael Grimm wrote:
    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a separate origin?

    +1

    +1 - and then depending packages can decide how they can migrate, without breaking user setups when running a naive "pkg upgrade"... (POLA).

    gert
    --
    "If was one thing all people took for granted, was conviction that if you
    feed honest figures into a computer, honest figures come out. Never doubted
    it myself till I met a computer with a sense of humor."
    Robert A. Heinlein, The Moon is a Harsh Mistress

    Gert Doering - Munich, Germany gert@greenie.muc.de


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Kurt Jaeger@pi@freebsd.org to muc.lists.freebsd.ports on Sat Jul 4 11:47:54 2026
    From Newsgroup: muc.lists.freebsd.ports

    Hi!

    Before upgrading, export all existing databases using the old v0.9
    mdb_dump utility, then import them with the new v1.0 mdb_load after
    upgrading. There is no support for opening v0.9 database files
    directly with LMDB 1.0.

    Example migration procedure:
    # mdb_dump -a /path/to/db > /tmp/mydb.dump
    # pkg upgrade databases/lmdb
    # mdb_load -f /tmp/mydb.dump /path/to/newdb
    <snip>

    One can probably extract the old mdb_dump from the old package under
    some directory and use it after the upgrade as well ?

    Has someone tested this ?
    --
    pi@FreeBSD.org +49 171 3101372 Now what ?


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Daniel Engberg@daniel.engberg.lists@pyret.net to muc.lists.freebsd.ports on Sat Jul 4 12:44:09 2026
    From Newsgroup: muc.lists.freebsd.ports

    This is a multi-part message in MIME format. --------------LOzR9OtimUPu10pMvEsBMRSl
    Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit

    On 2026-07-04 11:45, Gert Doering wrote:
    Hi,

    On Sat, Jul 04, 2026 at 11:40:32AM +0200, Michael Grimm wrote:
    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a separate origin?
    +1
    +1 - and then depending packages can decide how they can migrate, without breaking user setups when running a naive "pkg upgrade"... (POLA).

    gert

    There are two different issues here, Migration and compatibility People
    really need to get rid of the "I don't want to change" mindset
    especially tracking head/master. POLA is not about that, we can't police upstream projects and doing forks / extensive patching isn't reasonable
    and neither what any other distro / package repo does. If upstream makes changes users need to follow. We can apply POLA to projects we own but expecting that to carry out to third party is obviously out of our
    hands. What we can do is however to inform about breaking changes and
    try to make the migration as painless as possible. I'm not suggesting
    that we knowingly should break things just to the sake of it but on the
    other hand these kind of issues have occurred before and will occur from
    time to time. Compatibility As Arch people have noted there also seems
    to be runtime and (?) build issues, that however needs to be addressed.
    Given that time frame and affected ports it's likely a better idea to
    revert the commit and create a meta PR for upgrading 1.x instead of
    having multiple versions in parallel.

    What can we learn from this? Changelogs are important and to be fair one
    could argue that this information wasn't easily obtainable / well communicated.

    https://lists.openldap.org/hyperkitty/list/openldap-announce@openldap.org/thread/P347LVJF4UAA3XL7D7ITJGJI54UAARXC/
    https://www.lmdb.tech --> "Technical Information" (below download
    button) --> More documentation --> "Upgrading From Release 0.9"http://www.lmdb.tech/doc/upgrading.html

    No CI pre commit doesn't help, "exp-run"-ish requirement/checks and
    expecting a committer to have access to two different releases and at
    least three arch's (amd64, aarch64 and if possible i386 (I know its
    being phased out)) is a bit of a tall order, we're all humans and we can
    miss things. Best regards, Daniel

    --------------LOzR9OtimUPu10pMvEsBMRSl
    Content-Type: text/html; charset=UTF-8
    Content-Transfer-Encoding: 7bit

    <!DOCTYPE html>
    <html>
    <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
    </head>
    <body>
    <div class="moz-cite-prefix">On 2026-07-04 11:45, Gert Doering
    wrote:<br>
    </div>
    <blockquote type="cite" cite="mid:akjWRlV1VJzDWDRb@greenie.muc.de">
    <pre wrap="" class="moz-quote-pre">Hi,

    On Sat, Jul 04, 2026 at 11:40:32AM +0200, Michael Grimm wrote:
    </pre>
    <blockquote type="cite">
    <blockquote type="cite">
    <pre wrap="" class="moz-quote-pre">Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a separate origin?
    </pre>
    </blockquote>
    <pre wrap="" class="moz-quote-pre">
    +1
    </pre>
    </blockquote>
    <pre wrap="" class="moz-quote-pre">
    +1 - and then depending packages can decide how they can migrate, without breaking user setups when running a naive "pkg upgrade"... (POLA).

    gert
    </pre>
    </blockquote>
    <p><span style="white-space: pre-wrap">There are two different issues here,

    Migration and compatibility

    People really need to get rid of the "I don't want to change" mindset especially tracking head/master. </span><span
    style="white-space: pre-wrap">POLA is not about that, we can't police upstream projects and doing forks / extensive patching isn't reasonable and neither what any other distro / package repo does. </span><span
    style="white-space: pre-wrap">If upstream makes changes users need to follow. We can apply POLA to projects we own but expecting that to carry out to third party is obviously out of our hands. What we can do is however to inform about breaking changes and try to make the migration as painless as possible. I'm not suggesting that we knowingly should break things just to the sake of it but on the other hand these kind of issues have occurred before and will occur from time to time.

    Compatibility
    As Arch people have noted there also seems to be runtime and (?) build issues, that however needs to be addressed. Given that time frame and affected ports it's likely a better idea to revert the commit and create a meta PR for upgrading 1.x instead of having multiple versions in parallel.</span></p>
    <p><span style="white-space: pre-wrap">What can we learn from this? Changelogs are important and to be fair one could argue that this information wasn't easily obtainable / well communicated.
    </span></p>
    <p><span style="white-space: pre-wrap"><a class="moz-txt-link-freetext" href="https://lists.openldap.org/hyperkitty/list/openldap-announce@openldap.org/thread/P347LVJF4UAA3XL7D7ITJGJI54UAARXC/">https://lists.openldap.org/hyperkitty/list/openldap-announce@openldap.org/thread/P347LVJF4UAA3XL7D7ITJGJI54UAARXC/</a>

    <a class="moz-txt-link-freetext" href="https://www.lmdb.tech">https://www.lmdb.tech</a> --&gt; "Technical Information" (below download button) --&gt; More documentation --&gt; "Upgrading From Release 0.9"</span><span
    style="white-space: pre-wrap">
    <a class="moz-txt-link-freetext" href="http://www.lmdb.tech/doc/upgrading.html">http://www.lmdb.tech/doc/upgrading.html</a></span></p>
    <p><span style="white-space: pre-wrap">No CI pre commit </span><span
    style="white-space: pre-wrap">doesn't help</span><span
    style="white-space: pre-wrap">, "exp-run"-ish requirement/checks and expecting a committer to have access to two different releases and at least three arch's (amd64, aarch64 and if possible </span><span
    style="white-space: pre-wrap">i386 (I know its being phased out)</span><span
    style="white-space: pre-wrap"> ) is a bit of a tall order, we're all humans and we can miss things.

    Best regards,
    Daniel</span></p>
    </body>
    </html>

    --------------LOzR9OtimUPu10pMvEsBMRSl--


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Tomoaki AOKI@junchoon@dec.sakura.ne.jp to muc.lists.freebsd.ports on Sat Jul 4 22:31:42 2026
    From Newsgroup: muc.lists.freebsd.ports

    On Sat, 4 Jul 2026 12:44:09 +0200
    Daniel Engberg <daniel.engberg.lists@pyret.net> wrote:

    On 2026-07-04 11:45, Gert Doering wrote:
    Hi,

    On Sat, Jul 04, 2026 at 11:40:32AM +0200, Michael Grimm wrote:
    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a separate origin?
    +1
    +1 - and then depending packages can decide how they can migrate, without breaking user setups when running a naive "pkg upgrade"... (POLA).

    gert

    There are two different issues here, Migration and compatibility People really need to get rid of the "I don't want to change" mindset
    especially tracking head/master. POLA is not about that, we can't police upstream projects and doing forks / extensive patching isn't reasonable
    and neither what any other distro / package repo does. If upstream makes changes users need to follow. We can apply POLA to projects we own but expecting that to carry out to third party is obviously out of our
    hands. What we can do is however to inform about breaking changes and
    try to make the migration as painless as possible. I'm not suggesting
    that we knowingly should break things just to the sake of it but on the other hand these kind of issues have occurred before and will occur from time to time. Compatibility As Arch people have noted there also seems
    to be runtime and (?) build issues, that however needs to be addressed. Given that time frame and affected ports it's likely a better idea to
    revert the commit and create a meta PR for upgrading 1.x instead of
    having multiple versions in parallel.

    What can we learn from this? Changelogs are important and to be fair one could argue that this information wasn't easily obtainable / well communicated.

    https://lists.openldap.org/hyperkitty/list/openldap-announce@openldap.org/thread/P347LVJF4UAA3XL7D7ITJGJI54UAARXC/
    https://www.lmdb.tech --> "Technical Information" (below download
    button) --> More documentation --> "Upgrading From Release 0.9"http://www.lmdb.tech/doc/upgrading.html

    No CI pre commit doesn't help, "exp-run"-ish requirement/checks and expecting a committer to have access to two different releases and at
    least three arch's (amd64, aarch64 and if possible i386 (I know its
    being phased out)) is a bit of a tall order, we're all humans and we can miss things. Best regards, Daniel

    Hi.

    Don't forget, databases/lmdb is a "database" port.

    For library incompatibilities, basically it can be worked around
    by ports side, not bothering users (excepts waiting for new pkg
    to be available). With this aspect, nothing different with
    non-database ports.

    For migrations, the "better" way would be to allow installing and
    running new one aside of previous one, run both at the same time, and
    vacuum from previous and pass the data to new one with special
    "migraiton" mode.

    (Off topic here, but it would be nice if different versions of databases/postgres* ports don't conflict each other.)


    Of course, the best is to versioning database format and new
    software versions always support reading older format, and provide
    converters for painless (except disk capacity) migrations.
    So I've noted "better" above.

    Regards.
    --
    Tomoaki AOKI <junchoon@dec.sakura.ne.jp>


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jordan M@freebsdlists@montesse.ca to muc.lists.freebsd.ports on Sat Jul 4 10:04:33 2026
    From Newsgroup: muc.lists.freebsd.ports

    On 2026-07-04 2:40, Michael Grimm wrote:
    Jordan M <freebsdlists@montesse.ca> wrote:

    Following up from an issue raised over with Arch Linux[1] about
    postfix, knot, bogofilter, et al not working with lmdb 1.0.0, I tested
    on FreeBSD and confirmed lmdb-1.0.0 silently breaks consumers that work
    with 0.93.5 in a way not fixed by rebuilding after a library version bump.

    Now, I found in UPDATING (should have looked earlier):

    <snip>
    20260702:
    AFFECTS: users of databases/lmdb
    [...]
    That would be a whole lotta work for my numerous knot3 and
    postfix-current lmdb databases at some servers ...
    If the files used by the port are the same format that mdb_dump can
    read. AFAICT that's not the case for postfix. I couldn't get mdb_dump
    to recognize the postmap-generated file as a valid database.


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Michael Grimm@trashcan@ellael.org to muc.lists.freebsd.ports on Sat Jul 4 20:37:45 2026
    From Newsgroup: muc.lists.freebsd.ports

    Jordan M <freebsdlists@montesse.ca> wrote:
    On 2026-07-04 2:40, Michael Grimm wrote:
    That would be a whole lotta work for my numerous knot3 and
    postfix-current lmdb databases at some servers ...
    If the files used by the port are the same format that mdb_dump can read. AFAICT that's not the case for postfix. I couldn't get mdb_dump to recognize the postmap-generated file as a valid database.
    IMHO you don't need mdb_dump/mdb_load w.r.t. to postfix' lmdb databases. It should be sufficient to re-run 'postmap lmdb:YOUR_PLAIN_TABLE_OR_WHATSOEVER' on all relevant plain text files you need to convert to lmdb databases. (To be on the save side I would delete that YOUR_PLAIN_TABLE_OR_WHATSOEVER.lmdb beforehand.)
    But: I haven't done that, yet.
    Regards,
    Michael
    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jordan M@freebsdlists@montesse.ca to muc.lists.freebsd.ports on Sat Jul 4 12:46:53 2026
    From Newsgroup: muc.lists.freebsd.ports

    On 2026-07-04 11:37, Michael Grimm wrote:
    Jordan M <freebsdlists@montesse.ca> wrote:
    If the files used by the port are the same format that mdb_dump
    can read. AFAICT that's not the case for postfix. I couldn't get
    mdb_dump to recognize the postmap-generated file as a valid
    database.

    IMHO you don't need mdb_dump/mdb_load w.r.t. to postfix' lmdb
    databases.

    It should be sufficient to re-run 'postmap lmdb:YOUR_PLAIN_TABLE_OR_WHATSOEVER' on all relevant plain text
    files you need to convert to lmdb databases. (To be on the save side
    I would delete that YOUR_PLAIN_TABLE_OR_WHATSOEVER.lmdb beforehand.)

    Normally it would be, however:

    On 2026-07-03 16:47, Jordan M wrote:
    # pkg info -d postfix | grep lmdb
    lmdb-1.0.0,1
    lmdb-1.0.0,1 (liblmdb.so.1)
    # postmap lmdb:/tmp/testdb
    postmap: fatal: error updating lmdb:/tmp/testdb: MDB_BAD_TXN:
    Transaction must abort, has a child, or is invalid
    #

    That error occurred whether or not /tmp/testdb.lmdb already existed.


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Chris@portmaster@bsdforge.com to muc.lists.freebsd.ports on Sat Jul 4 18:20:55 2026
    From Newsgroup: muc.lists.freebsd.ports

    --=_5a01901855131e5b4337e886335fb179
    Content-Transfer-Encoding: 7bit
    Content-Type: text/plain; charset=US-ASCII;
    format=flowed

    On 2026-07-04 03:44, Daniel Engberg wrote:
    On 2026-07-04 11:45, Gert Doering wrote:
    Hi,

    On Sat, Jul 04, 2026 at 11:40:32AM +0200, Michael Grimm wrote:
    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a
    separate origin?
    +1
    +1 - and then depending packages can decide how they can migrate, without
    breaking user setups when running a naive "pkg upgrade"... (POLA).

    gert

    There are two different issues here, Migration and compatibility People really
    need to get rid of the "I don't want to change" mindset especially tracking head/master. POLA is not about that, we can't police upstream projects and doing
    forks / extensive patching isn't reasonable and neither what any other distro /
    package repo does. If upstream makes changes users need to follow. We can apply
    POLA to projects we own but expecting that to carry out to third party is obviously out of our hands. What we can do is however to inform about breaking
    changes and try to make the migration as painless as possible. I'm not suggesting
    that we knowingly should break things just to the sake of it but on the other hand
    these kind of issues have occurred before and will occur from time to time. Compatibility As Arch people have noted there also seems to be runtime and (?)
    build issues, that however needs to be addressed. Given that time frame and affected ports it's likely a better idea to revert the commit and create a meta PR
    for upgrading 1.x instead of having multiple versions in parallel.

    What can we learn from this? Changelogs are important and to be fair one could
    argue that this information wasn't easily obtainable / well communicated.

    https://lists.openldap.org/hyperkitty/list/openldap-announce@openldap.org/thread/P347LVJF4UAA3XL7D7ITJGJI54UAARXC/
    https://www.lmdb.tech --> "Technical Information" (below download button)
    More
    documentation --> "Upgrading From Release 0.9"http://www.lmdb.tech/doc/upgrading.html

    No CI pre commit doesn't help, "exp-run"-ish requirement/checks and expecting a
    committer to have access to two different releases and at least three arch's (amd64, aarch64 and if possible i386 (I know its being phased out)) is a bit of a
    tall order, we're all humans and we can miss things. Best regards, Daniel

    I don't understand the long winded defense of the way this was done. IMDB 0.9 and 1.0
    are NOT mutually incompatible. IOW they can both live on the same system WITHOUT
    conflict. Their SO names are different. As such a port or maintainer can choose to
    which version they wish to link against. There was little to no reason to
    make
    v1.o mandatory. Kindly bring .9 back. Having done so in the first place would have
    made for a much quieter and uneventful transition.

    Thank you. :)

    --Chris

    --=_5a01901855131e5b4337e886335fb179
    Content-Transfer-Encoding: 7bit
    Content-Type: application/pgp-keys;
    name=0xE512722F.asc
    Content-Disposition: attachment;
    filename=0xE512722F.asc;
    size=3074

    -----BEGIN PGP PUBLIC KEY BLOCK-----

    mQENBGf/G0IBCADARuJc6IcwOe3jv7dQsP1X/EIHvCFExPbTmlMNFMXbMMccQUnV o8ayEn+wmTvPhw7uL3PDk7DQs16W1sN2b8UMFc804cVWNGtoG3rA+Np+TFEYlXJx eh5Q42VHptkuwzHKl+q2utkpRlS7uHyfjsInQAoHxLyi/wrsaZTHHhDbLLhJ5Ez0 arohQ2Q1w0M5e9rW8Fy5rpC7RpC6uO1SZMxcbdqURI/BBqxbiD1iW62cDWFkfFX+ dtaEXghFV7BIBMDSrgIunGoEfdMZgXys7O6bPWn8z0cuOZIPj4HrjoCYARyQ+sdc rjz/k06SLM/UvEZDorJhT4DbYrwMNvaPWJiPABEBAAG0HkNocmlzIDxic2QtbGlz dHNAYnNkZm9yZ2UuY29tPokBNQQQAQgAHwUCZ/8bQgYLCQcIAwIEFQgKAgMWAgEC GQECGwMCHgEACgkQVKBqaOUSci8bSwf/fK3QcTYXRMrv82HIp4SiGCSD7/bRmyWr ipv2vzknGFHxPBN4AEWIqF/U4j5oDXaodyU6xsy59Z47/lgbyzyZiVR6nmJVgZVf el/EgwnLt7ZuYGLLEhIN2pd9itJkB8PMPZrUHMWgIw8BxX5YFYGuyiNe9pGn0Coj 98t/v3fouhqksH+BpB4TBHJBBDSxSiMm66VTJX4Xcnpf0ZnQVP4GBuoyodnFBfdI wqftPLESsCC08lUhD2j7v2NRWwMi/q3ed8D6VCKPImBByYnBZL5gu56K5bwqaQfN itu06APuIYnG71qxgn1EPO63lovWP5NZGgOKvzs3K+JfPF79BiOUFbQjQ2hyaXMg PG1haWxvcEBocmNvbW11bmljYXRpb25zLm5ldD6JATEEEAEIABwFAmf/G0IGCwkH CAMCBBUICgIDFgIBAhsDAh4BAAoJEFSgamjlEnIvBH8H9RGwzZuU6+zvH1WjQa97 yWpEt9rC+BIBJThev2Cpls2LqBqIeIQVZPnyLAZWgFaiezL6+xbvcNt6OnfidIYa x8iRwCMC6/Bs8H2Wef9qfGxXi+jHPLYQk3juiZVmBhIK6FJZkzaW4wSiawofwzbp zqNxO8dZ0j4foaJZrNi8iqsvKjiiHoSFaJtumIThAeydI18CNLeFaS53sk5nad6I wCYeFKmJ/22dMP7DOFEgyG1iNYgY+AGREMkEsBiLpqYjJ5asK+1UdUy/TRly1hOt HHxCiX0Fh9ZYM2vLIj7sq4LKaMPGeYC3qTqBYugVeyz7LkiI2ft/BKveA5JxuYKk ZrQiQ2hyaXMgPG5hbm9nQGhyY29tbXVuaWNhdGlvbnMubmV0PokBMgQQAQgAHAUC Z/8bQgYLCQcIAwIEFQgKAgMWAgECGwMCHgEACgkQVKBqaOUSci+4Bwf8D0Ogk2/X ud/CsAgHozwzKPqfesL5SRWM14hLnU9/EHoplnZgNexbVY1wXIi2FYPo5cve9QxW Nmt3S3UTF9j2fGqv0wmeHv3EqogFUHnftLyWpbeTPOFDMIQp/BOD6ygfeXxXWxRT L6zvUkSrDtHvkQHPWGRxwP+ihWjpw9AQR/R4/qAuTAZZM0O7UnJEo4mWXatl+utF wegG2giwFTTxfF+1rMpFtUDjYCpRQ6ZmE+gC1mHUMoH7GJMQv12DbqwKrxtwGfd0 AJNO3ZDnxl24BmIfl1YqQGZQ5iIH7At4YItESbU45hoNNsG9oDrsil78EUCAtXHd UPScj+eXaeAkgrQfQ2hyaXMgPHBvcnRtYXN0ZXJAYnNkZm9yZ2UuY29tPokBMgQQ AQgAHAUCZ/8bQgYLCQcIAwIEFQgKAgMWAgECGwMCHgEACgkQVKBqaOUSci9o7Af+ Lwu5hJlI5HZNGwAll7QTIFZVW+y4OEg+amhxTDGbAAqlnSIkHC1KgkmIOOrThme3 kTFCqfIIsuP73yKxHq6kRG0zH5/7asAPNAUOfzD7B2o/gMyuTRKyG5r9f3UmACr4 6qvtFhIwROXr6+NNT2IKg3l0/8F58A0N/TR8D2PTHeo4x6jYcZQDCrCy7BAdk3cu V16k4z/1UzRa07b5McezbWL20cIaZ+dqNcCjKZpzPlTyTCGgrNNtaDpNVhoWUKMB YNcKql+tfC1IpX8l+IU6OBKcDKMkQojvO1QrZqY8MDJGo8jq/CtotQ8+IpAai3Bx dQEsxrxlcKTR4rUqvd8VGbkBDQRn/xtCAQgAv5Nv/aQN72xsLik+K73PJwpUmyhu vnI6stM6dSecylXVHjZ7C4n/m0eQEeQCl+9lByHR9N8H+WS3DtAd4pmciiIxRQLA JZiuaLYcy9ziy1h7130VoR7hhJHzo9FIhWkTGlCDX3egUZrYhMiwFUO8lNltLB8o TBvIrMSsnUzawtQjq/otv0Jf+oBPbG+gIYnAm7w6r86n/l+eVxf5eEoS7wV0DJfp b2jE5zWErWk8I/tq4e8T+1VQeVQR6wz+NrUCSxkPkpNAm19AFUHOk//yvMGWVlDW F6gr3ErN2a0w/kZ0lz3Msxsb87QT+MnJf/T3cuEqdTIoSk74BfNEAdMohQARAQAB iQEfBBgBCAAJBQJn/xtCAhsMAAoJEFSgamjlEnIvyvIH/26zytSVNDaxtprg7XtX LerIWf9RyVx8omCw/lXKRCcgkfwD7QR+nSZ0thWOGMpcnivjuReeVRkz/webUF47 BXJ/Tge07nrxdtyTIHBbp35fPIriaKaII6YWc2Ufdxwv+cD8PADS6gQWAlgrWLmn VmYtyHs4kwtiPZyUyuBdWnZal2GyYY0WVwYjvbk95eInwOaIdoTjesJ7ZhUFu155 r4hh9GlvM0uv8WJ5Mw9wvHa5fIM205I5g0IWC7yvTwwwKHlV4JQQOqMwfv569OEl 1GKqA12nSVziB1+UV+I0NqOABWi/MOi+IySPzYP+XgdPfRNx4vmoHYZwWOQ3t4Jd
    TEM=
    =oj6y
    -----END PGP PUBLIC KEY BLOCK-----

    --=_5a01901855131e5b4337e886335fb179--


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Michael Grimm@trashcan@ellael.org to muc.lists.freebsd.ports on Sun Jul 5 09:08:49 2026
    From Newsgroup: muc.lists.freebsd.ports

    Jordan M <freebsdlists@montesse.ca> wrote:
    On 2026-07-04 11:37, Michael Grimm wrote:
    Jordan M <freebsdlists@montesse.ca> wrote:
    If the files used by the port are the same format that mdb_dump
    can read. AFAICT that's not the case for postfix. I couldn't get mdb_dump to recognize the postmap-generated file as a valid
    database.
    IMHO you don't need mdb_dump/mdb_load w.r.t. to postfix' lmdb
    databases.
    It should be sufficient to re-run 'postmap lmdb:YOUR_PLAIN_TABLE_OR_WHATSOEVER' on all relevant plain text files you need to convert to lmdb databases. (To be on the save side
    I would delete that YOUR_PLAIN_TABLE_OR_WHATSOEVER.lmdb beforehand.)

    Normally it would be, however:

    On 2026-07-03 16:47, Jordan M wrote:
    # pkg info -d postfix | grep lmdb
    lmdb-1.0.0,1
    lmdb-1.0.0,1 (liblmdb.so.1)
    # postmap lmdb:/tmp/testdb
    postmap: fatal: error updating lmdb:/tmp/testdb: MDB_BAD_TXN:
    Transaction must abort, has a child, or is invalid
    #

    That error occurred whether or not /tmp/testdb.lmdb already existed.
    I used a test jail and can reproduce this issue:
    test> ldd /usr/local/lib/postfix/* | grep lmdb /usr/local/lib/postfix/postfix-lmdb.so:
    liblmdb.so.1 => /usr/local/lib/liblmdb.so.1 (0x90234ff3000)
    test> la /usr/local/lib/liblmdb*
    -rw-r--r-- 1 root wheel uarch 137992 Jul 5 08:51 /usr/local/lib/liblmdb.a lrwxr-xr-x 1 root wheel uarch 14 Jul 5 08:51 /usr/local/lib/liblmdb.so -> liblmdb.so.1.0
    lrwxr-xr-x 1 root wheel uarch 14 Jul 5 08:51 /usr/local/lib/liblmdb.so.1 -> liblmdb.so.1.0
    -rw-r--r-- 1 root wheel uarch 114040 Jul 5 08:51 /usr/local/lib/liblmdb.so.1.0
    test> postmap lmdb:TEST
    postmap: fatal: error updating lmdb:TEST: MDB_BAD_TXN: Transaction must abort, has a child, or is invalid
    Well, I will stick with lmdb 0.9.35.
    Regards,
    Michael
    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From David Wolfskill@david@catwhisker.org to muc.lists.freebsd.ports on Sun Jul 5 05:10:11 2026
    From Newsgroup: muc.lists.freebsd.ports

    --uZavpLgsSakHSY4j
    Content-Type: text/plain; charset=iso-8859-1
    Content-Disposition: inline
    Content-Transfer-Encoding: quoted-printable

    On Fri, Jul 03, 2026 at 11:59:27PM +0000, Einar Bjarni Halld=F3rsson wrote:
    On 7/3/26 11:47 PM, Jordan M wrote:
    ...
    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a
    separate origin?
    =20
    =20
    I agree with this. Upgrading lmdb broke both knot3 and bind920 for me. I'=
    ve
    fixed it by downgrading lmdb in a ports-overlay tree that poudriere uses.
    =20
    .einar

    For my use cases, the update to lmdb affected mail/mutt and dns/bind920.

    However -- in my use cases -- neither was broken.

    I suspect that for mutt, the use is for ephemeral caches. (And in any
    case, there was an update to mutt within the last 24 hours.) And for
    bind920, I do not happen to make use of features that actually use lmdb. (Thanks for the "heads up," though, so I spent a few minutes scanning for
    the existence of any possible lmdb files ... and then updated rather
    more cautiously than ususal, with a few extra checking steps along the
    way.)

    Peace,
    david
    --=20
    David H. Wolfskill david@catwhisker.org

    See https://www.catwhisker.org/~david/publickey.gpg for my public key.

    --uZavpLgsSakHSY4j
    Content-Type: application/pgp-signature; name=signature.asc

    -----BEGIN PGP SIGNATURE-----

    iPEEARYKAJkWIQRCec5RsK7Enudh3yGB9MJ9AwUELQUCakpJoxsUgAAAAAAEAA5t YW51MiwyLjUrMS4xMiwyLDNfFIAAAAAALgAoaXNzdWVyLWZwckBub3RhdGlvbnMu b3BlbnBncC5maWZ0aGhvcnNlbWFuLm5ldDQyNzlDRTUxQjBBRUM0OUVFNzYxREYy MTgxRjRDMjdEMDMwNTA0MkQACgkQgfTCfQMFBC2UlgEAyt7NF5BxTLv6qtTL2KjB RbmPLaWJD7eC1Dp7MgRPHKMA/Rn6bsdBOxH/1vCz47MBXFht2U2UPc916wBsrlO9
    BxYK
    =JbTp
    -----END PGP SIGNATURE-----

    --uZavpLgsSakHSY4j--


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From A FreeBSD User@freebsd@walstatt-de.de to muc.lists.freebsd.ports on Sun Jul 19 15:53:12 2026
    From Newsgroup: muc.lists.freebsd.ports

    --Sig_/HT6yUnG.g=d5V9+SZoVZvpl
    Content-Type: text/plain; charset=UTF-8
    Content-Transfer-Encoding: quoted-printable

    Am Tage des Herren Sun, 5 Jul 2026 05:10:11 -0700
    David Wolfskill <david@catwhisker.org> schrieb:

    On Fri, Jul 03, 2026 at 11:59:27PM +0000, Einar Bjarni Halld=C3=B3rsson w=
    rote:
    On 7/3/26 11:47 PM, Jordan M wrote:
    ... =20
    Can we get databases/lmdb reverted to 0.93.5, and 1.0.0 added as a separate origin?
    =20
    =20
    I agree with this. Upgrading lmdb broke both knot3 and bind920 for me. =
    I've
    fixed it by downgrading lmdb in a ports-overlay tree that poudriere use=
    s.
    =20
    .einar =20
    =20
    For my use cases, the update to lmdb affected mail/mutt and dns/bind920.

    I had some issues after the upgrade to lmdb 1.0.0, but all resolved after r= ecompiling theport
    recusively, especially bind920. After that, everything worked as expected. = Same with
    poudriere. I didn't see any miscompilation due to lmdb 1.0.0 - in my use ca= ses and
    restricteted set of ports compiled via poudriere!

    Reverting to 0.93.X made the whole process going backwards and really messy=
    ! Wasn't necessary,
    I presume.

    =20
    However -- in my use cases -- neither was broken.
    =20
    I suspect that for mutt, the use is for ephemeral caches. (And in any
    case, there was an update to mutt within the last 24 hours.) And for bind920, I do not happen to make use of features that actually use lmdb. (Thanks for the "heads up," though, so I spent a few minutes scanning for
    the existence of any possible lmdb files ... and then updated rather
    more cautiously than ususal, with a few extra checking steps along the
    way.)
    =20
    Peace,
    david



    --=20

    A FreeBSD user

    --Sig_/HT6yUnG.g=d5V9+SZoVZvpl
    Content-Type: application/pgp-signature
    Content-Description: OpenPGP digital signature

    -----BEGIN PGP SIGNATURE-----

    iJEEARYKADkWIQRQheDybVktG5eW/1Kxzvs8OqokrwUCalzW4xsUgAAAAAAEAA5t YW51MiwyLjUrMS4xMiwyLDMACgkQsc77PDqqJK82/QEAxnZnBMIIzrsfePvchcl4 0kH1MSuOnFRyDlR/7SNwb1ABALVzRK0+4hSg9uBE0tGy1eUUAOyMYZ6UpqWsSyFi
    HakO
    =dGjx
    -----END PGP SIGNATURE-----

    --Sig_/HT6yUnG.g=d5V9+SZoVZvpl--


    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From =?utf-8?Q?Dag-Erling_Sm=C3=B8rgrav?=@des@FreeBSD.org to muc.lists.freebsd.ports on Tue Jul 21 10:25:59 2026
    From Newsgroup: muc.lists.freebsd.ports

    A FreeBSD User <freebsd@walstatt-de.de> writes:
    Reverting to 0.93.X made the whole process going backwards and really
    messy! Wasn't necessary, I presume.
    It most certainly was, as lmdb 1.0.0 is not backward compatible with
    0.9.x, so upgrading requires manual intervention to convert existing
    database files to the new format. That is a really unfortunate property
    for an embedded database commonly used to store metadata the user
    doesn't even realize exists.
    DES
    --
    Dag-Erling Sm|+rgrav - des@FreeBSD.org
    --
    Posted automagically by a mail2news gateway at muc.de e.V.
    Please direct questions, flames, donations, etc. to news-admin@muc.de
    --- Synchronet 3.22a-Linux NewsLink 1.2