From mboxrd@z Thu Jan 1 00:00:00 1970 From: Adolf Belka To: development@lists.ipfire.org Subject: Re: Problem updating python3-cryptography Date: Thu, 16 Jun 2022 15:43:19 +0200 Message-ID: In-Reply-To: <8EDE17AA-E05C-47C0-BBCF-52B1EC5565ED@ipfire.org> MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="===============9026934101826683354==" List-Id: --===============9026934101826683354== Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable Hi Michael, On 16/06/2022 12:45, Michael Tremer wrote: > Hello Adolf, >=20 > I ran into those problems too when I packaged the first bunch of rust packa= ges. >=20 > The problem is, that cargo cannot easily differentiate what dependencies it= actually needs. Developers of those crates can list certain dependencies for= only certain platforms (e.g. Windows), and those will only be built on those= platforms then. However, they still need to be all present on all systems - = even though they are not being built. >=20 > Not the best design, but probably okay for everyone who downloads those pac= kages live using cargo. >=20 > What I have done in those cases is just remove those dependencies manually: >=20 > https://git.ipfire.org/?p=3Dipfire-2.x.git;a=3Dblob;f=3Dsrc/patches/rust-di= rs-1.0.5-fix-metadata.patch;h=3D0c51222127924cab1d6a85ee099c7bd05667948a;hb= =3DHEAD > https://git.ipfire.org/?p=3Dipfire-2.x.git;a=3Dblob;f=3Dsrc/patches/rust-ge= trandom-0.2.4-fix-metadata.patch;h=3D083b6d6c80a5577a349f8dbe34929b48229ca05d= ;hb=3DHEAD > https://git.ipfire.org/?p=3Dipfire-2.x.git;a=3Dblob;f=3Dsrc/patches/rust-ho= me-0.5.3-fix-metadata.patch;h=3D78e154af69554b3b53ab067d8f1722ef069ddf7d;hb= =3DHEAD >=20 > There are lots more in src/patches. >=20 > And this is the only way to avoid packaging stuff that will never and can n= ever be used on IPFire. >=20 > Hope this helps. That indeed helped. I now know how to get rid of them. Have created the patch= to exclude winapi which will stop the need for the other winapi files. The build is running now. I think I am now close to being able to release the updated python3-cryptogra= phy. Thanks very much for the help. I have learnt a lot with this. Regards, Adolf. >=20 > -Michael >=20 >> On 15 Jun 2022, at 21:02, Adolf Belka wrote: >> >> Hi Michael, >> >> I added the rust-indoc and rust-paste packages as explicitly defined versi= ons to deal with the requirement for an older version. After another couple o= f changes I have been able to get python3-cryptography to build successfully. >> >> However, the build ended up requiring three packages that seem totally red= undant to me as they are related to windows but I can't figure out how to eli= minate them. Do these have to stay being included or is there a way to stop t= hem being required. >> >> The packages are:- >> >> rust-winapi >> rust-winapi-i686-pc-windows-gnu >> rust-winapi-x86_64-pc-windows-gnu >> >> Regards, >> >> Adolf >> >> >> On 03/05/2022 23:02, Adolf Belka wrote: >>> Hi Michael, >>> >>> On 03/05/2022 12:32, Michael Tremer wrote: >>>> Hello Adolf, >>>> >>>> It looks like it is looking for a Rust package called asn1. >>>> >>>> You can create this in a very simple way by running: >>>> >>>> tools/download-rust-crate asn1 >>>> >>>> That will create a new LFS file for this package and automatically downl= oad any sources. You will only need to add this to make.sh and build it. >>> That is a very nice script. Worked very well. >>>> >>>> You might need some further dependencies later on. >>> That is an understatement. >>> I am at the stage where python3-cryptography has required asn1, ouroboros= and pem and each of then has had additional dependencies. so I have reached = an additional 20 new rust packages. 5 of those had to be separately re-downlo= aded as specific versions were required and not the latest version. >>> >>> Then python3-cryptography indicated that a newer rust-pyo3 was required b= ut not the most up to date version. I added that version in and then rust-pyo= 3 flagged up that the requirement was for indoc to be at version 0.3.6 but th= e currently installed version is 1.0.3. That would mean reverting indoc to a = much older version than the one already installed and running. The log file f= or this is below. >>> >>> I am not sure what to do now. >>> >>> Regards, >>> Adolf. >>> >>> >>> Building rust-pyo3 log >>> >>> May 3 17:30:37: Building rust-pyo3 pyo3-0.15.1.tar.gz checksum OK >>> + cd /usr/src/lfs >>> + make -f rust-pyo3 LFS_BASEDIR=3D/usr/src install >>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D Installing pyo3-0.15.1 ... >>> Install started; saving file list to /usr/src/lsalr ... >>> cd /usr/src/pyo3-0.15.1 && mkdir -p /usr/src/pyo3-0.15.1/.cargo && ec= ho "${CARGO_CONFIG}" > /usr/src/pyo3-0.15.1/.cargo/config && rm -f Cargo.lock >>> cd /usr/src/pyo3-0.15.1 && CARGOPATH=3D/usr/src/pyo3-0.15.1/.cargo RUSTC_= BOOTSTRAP=3D1 cargo --offline build --release -Z avoid-dev-deps -j6 >>> error: failed to select a version for the requirement `indoc =3D "^0.3.6"` >>> candidate versions found which didn't match: 1.0.3 >>> location searched: directory source `/usr/share/cargo/registry` (which is= replacing registry `crates-io`) >>> required by package `pyo3 v0.15.1 (/usr/src/pyo3-0.15.1)` >>> perhaps a crate was updated and forgotten to be re-vendored? >>> As a reminder, you're using offline mode (--offline) which can sometimes = cause surprising resolution failures, if this error is too confusing you may = wish to retry without the offline flag. >>> make: *** [rust-pyo3:77: /usr/src/log/pyo3-0.15.1] Error 101 >>> >>> >>> >>>> >>>> -Michael >>>> >>>>> On 10 Apr 2022, at 12:32, Adolf Belka wrote: >>>>> >>>>> Hi All, >>>>> >>>>> >>>>> I have gone through and updated all python3 modules that I found with a= new version. Everything built fine except for python3-cryptography. >>>>> >>>>> I have gone through the log info and it seems to me, but I might be wro= ng, that wants pyasnl1 but it seems to be looking for asnl1. >>>>> >>>>> However I have not been able to figure out where it should be changed t= o correct this, if my interpretation is even right. >>>>> >>>>> >>>>> I would appreciate any guidance/suggestions of what to look for. >>>>> >>>>> >>>>> Log feedback for the build of cryptography is attached below. >>>>> >>>>> >>>>> Regards, >>>>> >>>>> Adolf. >>>>> <_build.ipfire.log> >>>> >=20 --===============9026934101826683354==--