> > Always use distfiles if possible. > Alrighty then, since this seems to be the sole response on the matter, I'll go the distfiles route.
-- Jason Liu On Tue, Jul 14, 2020 at 10:37 PM Joshua Root <[email protected]> wrote: > On 2020-7-15 11:53 , Jason Liu wrote: > > I have a question regarding how best to fetch the source code for my > > Blender port. Blender's source code is split up into four pieces, which > > are of course structured as submodules in GitHub. I know that I could > > use the 'git submodule update --init' technique to obtain the submodules > > during the fetch phase, as has been discussed previously: > > > > https://trac.macports.org/ticket/50708 > > > > However, Blender does something which many other projects on GitHub > > don't: the four parts of the source code are always tagged with an > > identical release version number, and are always pushed up to GitHub > > simultaneously when a new release gets tagged. This makes it relatively > > easy to have my portfile fetch the four tarballs as distfiles. And since > > MacPorts supports multiple distfiles, this also seems like a viable > option. > > > > The reason why I ask is that if I fetch the tarballs of the four parts, > > the amount of data downloaded totals around 53 MB. If I use 'git > > submodule', the amount of data downloaded totals around 959 MB. Yes, > > disk storage and network bandwidth are cheap these days, but that's > > still a pretty big difference. > > > > So... which method should I use? > > Always use distfiles if possible. > > - Josh >
