Do we need to have the GCC test suites into the repository ? Cannot we have a script that downloads, untars, and adds into the package tree automatically. It is very long to checkout and we do not need to modify it. -- Valentin David valentin@lrde.epita.fr
De Nazareth, Jesus <valentin@lrde.epita.fr> wrote
Do we need to have the GCC test suites into the repository ? Cannot we have a script that downloads, untars, and adds into the package tree automatically. It is very long to checkout and we do not need to modify it.
I am against downloading it because it would make us depend on the Epita's Internet access, which is not a good idea. A better idea would be to keep the tests in a .tar.bz2 in the repository, and extract them in bootstrap.sh, thus reducing the checkout time. Since the tests are never modified, we don't need the incremental differences for them. Also, it would allow us to keep only the subset of the GCC test-suite that we use (we don't need all the tests). -- Clement Vasseur -o) [ nitro :: EPITA CSI 2005 ] /\\ "Programming is about being lazy." _\_V
"Clement" == Clement Vasseur <nitro@epita.fr> writes:
De Nazareth, Jesus <valentin@lrde.epita.fr> wrote
Do we need to have the GCC test suites into the repository ? Cannot we have a script that downloads, untars, and adds into the package tree automatically. It is very long to checkout and we do not need to modify it.
I am against downloading it because it would make us depend on the Epita's Internet access, which is not a good idea. A better idea would be to keep the tests in a .tar.bz2 in the repository, and extract them in bootstrap.sh, thus reducing the checkout time. Since the tests are never modified, we don't need the incremental differences for them. Also, it would allow us to keep only the subset of the GCC test-suite that we use (we don't need all the tests).
Ain't this cost paid *once*, and only once? Just at the initial checkout.
participants (3)
-
Akim Demaille -
Clement Vasseur -
valentin＠lrde.epita.fr