Preserving old Compiler Explorer links in Phabricator archive

[Endilll]

As Matt Godbolt highlighted in his blog post, godbolt.org/g/ links has to be opened to be preserved beyond August 2025. Content hosted on GitHub should be fine, because he used GitHub API to find all such links. I contacted him regarding our Phab archive, and he suggested that we run something along the lines of rg godbolt.org/g > /tmp/all-godbolt-links.txt on the server hosting the archive, and hand over the file to him. Can someone with enough access do that?

CC @mattgodbolt @MaskRay @asl

[llvmbot]

@llvm/issue-subscribers-infrastructure

Author: Vlad Serebrennikov (Endilll)

As Matt Godbolt highlighted in his [blog post](https://xania.org/202505/compiler-explorer-urls-forever), `godbolt.org/g/` links has to be opened to be preserved beyond August 2025. Content hosted on GitHub should be fine, because he used GitHub API to find all such links. I contacted him regarding our Phab archive, and he suggested that we run something along the lines of `rg godbolt.org/g > /tmp/all-godbolt-links.txt` on th server hosting the archive. Can someone with enough access do that?

CC @mattgodbolt @MaskRay @akorobeynikov

[mattgodbolt]

Thanks!! The situation may not be as scary as it used to be as apparently the internet archive are in the process of saving all goo.gl links, BUT it would be ace to have any "godbolt.org/g" links so we can be 100% sure we've archived them.

Thanks everyone

[Endilll]

The situation may not be as scary as it used to be as apparently the internet archive are in the process of saving all goo.gl links

The last time I've heard about it, copyright holders are very eager to bring the Internet Archive down, so I'd prefer to ensure that our history is preserved on first-party infrastructure.

[mattgodbolt]

Absolutely! I will absolutely bring all the links in-house, but it means if we miss a few there's at least a possibility we can look them up from the archive :) And if I can get an exhaustive dump I'll look to import those too

[MaskRay]

I no longer have access to the reviews.llvm.org website. @rnk may be able to arrange for someone to run the rg command on the archive and extract all godbolt.org/g links.

[rnk]

I constructed this command, and it's running: rg -o -I 'http[^<>]*godbolt.org/g[^"'\'' <>]*' > ~/godbolt-links.txt We'll see.

[mattgodbolt]

Thanks... that might miss gcc.godbolt.org and others, if I'm reading the regex right?

Edit, I think we're good! sorry :)

[rnk]

Here are the results, sorted and uniqued:
godbolt-links-uniq.txt

The initial grep command produced a 4959 line file, but that reduced to 241 lines with | sort | uniq.

[mattgodbolt]

Wonderful! Thank you so much. I appreciate it very much and am actually slightly disappointed that our existing "crawler"/weblog search has actually got them all already.

$ uv run links import ~/Downloads/godbolt-links-uniq.txt 
Found 241 godbolt.org/g/ URLs in /home/matthew/Downloads/godbolt-links-uniq.txt
Total unique fragments: 241 (241 cached, 0 to process)

Summary:
  Total: 241 unique fragments
  Cached: 241
  Expanded: 0
  Failed: 0

But - this makes me feel a LOT better that we have the important links archived.

Thank you all so much for your help with this!

--matt