Build refactoring, all downloads take place before any build attempts…#76
Build refactoring, all downloads take place before any build attempts…#76clienthax wants to merge 4 commits into
Conversation
|
Why not use the |
Develop branch last touched 6 years ago? |
… and are cached where possible.
Correct, the functionality has been available for 6 years already now. 😺 |
|
Hi Marcus! It's cool to see you're still reviewing ps3dev :) BTW - you may have seen a few of us have recently been working with bucanero to bring the toolchain and its surrounding repos up to speed. These changes are part of that. I don't know if you're on Discord but we would welcome your input on it here. |
|
I think the develop branch sort of got forgotten because there was no actual development the last 6 years (only a couple of hotfixes when certain URL:s stopped working). 😄 I could take a look at merging the branch now since there seems to be interest in development again? 🤪 |
|
Yes, that would be great! I spent the last couple of days working on the CIs for the toolchain and the Would be super cool if you could try merging it across and we can try and replicate that behaviour with the Savannah links in the |
It looks like the develop branch still attempts to download the config updates instead of using the local automake? |
"The local automake", it it exists at all, could have arbitrarily old config files. I'm all for using a faster source for getting the config files, but if we're getting outdated versions it defeats the purpose of replacing them in the packages in the first place... |
The purpose of updating the config.guess etc, is to make sure the current system can build said targets right (mainly itself)? Savannah is so terribly unreliable that I can't even access it from my network and it breaks the build. |
|
I echo the concern about Savannah. What is the danger of an outdated config? If it is included with automake as a dependency but not the latest, then the question is whether this has the potential to cause any problems. |
|
Well, since now getting The danger of an outdated config is that it will not recognize the build host (which is not necessarily ppc64) and break the build (for no apparent reason since we are cross-compiling anyway). It happened a lot with RISCV when that was new, it's bound to happen again. With the merge of develop branch (which I haven't pushed yet, currently doing a test build), |
Id suggest including them in the repository as a default with a env flag gated updater for it for ci simplicity if you really want to go that route, something like USE_INBUILT_MAKECONFIG_FILES, if defined will not run the savannah update |
|
@zeldin Il rebase this after you merge the develop branch in, setting this as draft for now. |
|
@clienthax Ok. I just pushed the merge to master. This also increases the number of available libs by 9. 😄 Note that you will need to pull the I also now remembered another thing I was working on 6 years ago but which was never merged to the |
|
I also synced the merge back to the |
3 participants
… and are cached where possible.