Arch Linux User Repository

@gborzi Thanks for your swift response and the instructions how to get the checksum for the downloaded files. The first checksum has been wrong, the other two identical, so the file gmsh-4.4.1-source.tgz has been affected. I deleted and downloaded the file again and the first checksum changed (still wrong). On the third attempt it was correct and I could install the update. Very strange, but all of that happened from a coorporate network, so possibly there was something fishy with the company's proxy server and the file got corrupted. Just checked at home on another machine and it updated witout any issue. So never mind and thank you guys for helping!

@lahwaacz I don't expect that users run makepkg -g and fill the checksums. Following chilichiller message, I've checked that the checksum in the PKGBUILD for gmsh-4.4.1-source.tgz is correct. Then asked to see the output of the same command on chilichiller system, to see if he has downloaded a corrupted version of the source.

@gborzi Users are not expected to run makepkg -g when building the package, you as the packager should ensure that the correct checksums for the current version are present in the PKGBUILD. If users blindly used makepkg -g to build all their packages, there would be no point to have the checksums at all, because there would be no verification in the first place.

@chilichiller I've just checked the sha256sum and it works fine for me, i.e. after deleting the old gmsh-4.4.1-source.tgz i ran

makepkg -g ==> Retrieving sources... -> Downloading gmsh-4.4.1-source.tgz... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 12.6M 100 12.6M 0 0 309k 0 0:00:41 0:00:41 --:--:-- 95 -> Found gmsh.desktop -> Found gmsh.completion ==> Generating checksums for source files... sha256sums=('853c6438fc4e4b765206e66a514b09182c56377bb4b73f1d0d26eda7eb8af0dc' '43a8ca33ac917ee7196fdae305ff2c8cb9ae1072569ee546c0ce8ff580c966ae' '11605e97636a56cf51e445e65019526ee253bd2e0553fb71ba6d94488dcd34ef') what's your output?

For version 4.4.1 there seems to be a mismatch of the sha256sum, it doesn't pass the validity check.

@jedbrown I'm compiling gmsh with med=4.0.0. Have you checked if .med file work with gmsh?

FWIW, I removed the MED version restriction and got what appears to be a successful build (was able to read and write MED files).

diff --git i/PKGBUILD w/PKGBUILD index 0b09e9a..adc0cbb 100644 --- i/PKGBUILD +++ w/PKGBUILD @@ -8,7 +8,7 @@ arch=('x86_64') url="http://gmsh.info/" license=('custom') makedepends=('cmake' 'desktop-file-utils' 'sed' 'swig' 'fltk' 'lapack' - 'med=3.3.1' 'opencascade' 'cairo' 'texlive-core') + 'med' 'opencascade' 'cairo' 'texlive-core') options=(!emptydirs) source=("${url}src/${pkgname}-${pkgver}-source.tgz" gmsh.desktop gmsh.completion) sha256sums=('54a236f5708bc105d5b60ddb2b95ea7062537ccd2720860377994c1a9bb86429' @@ -41,7 +41,7 @@ build() { }

package_gmsh() { - depends=('fltk' 'lapack' 'med=3.3.1' 'opencascade' 'cairo') + depends=('fltk' 'lapack' 'med' 'opencascade' 'cairo') optdepends=('gmsh-docs: docs for gmsh' 'python2: for onelab.py' 'python: for onelab.py')

@greyltc Try med3.

The med=3.3.1 requirement makes this impossible to install.

@lahwaacz I know, but since this package is updated quite frequently upstream, and it can take a lot of time to recompile for people without powerful machines, I'll wait a while. If there won't be another upstream release in a short time I'll go with the pkgrel.