opentimsr not on CRAN anymore

[sneumann]

Hi, opentimsr was pulled from CRAN for some non-technical reason, what changes would be needed to comply with their guidelines ? Yours, Steffen

[MatteoLacki]

Hello,

Would it be possible for you to automate the setup directly from github with devtools::install_github function? I know it's a change requiring additional package, but I simply don't have time for the next 2 weeks to look into this. The main problem for them was that the license file did not upload (which was not checked by their build system) and so we have been called liars and thieves by some old CRAN maintainer. It all makes me even less keen to even put it back there now.

Best wishes,

Dr Mateusz Krzysztof Łącki

tel. +49 159 01681376

GitHub: MatteoLacki https://github.com/MatteoLacki

On Tue, Jun 13, 2023 at 10:54 AM Steffen Neumann @.***> wrote:

Hi, opentimsr was pulled from CRAN for some non-technical reason, what changes would be needed to comply with their guidelines ? Yours, Steffen

— Reply to this email directly, view it on GitHub https://github.com/michalsta/opentims/issues/20, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA6H2ACMDEFDWHRZFNMT6BTXLATD7ANCNFSM6AAAAAAZEQVMYI . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[sneumann]

Thanks for the swift reply, yes, alternatively we could do install_github(), but something like CRAN or BioC still would make sense. Getting the LICENSE right is important, and there are certainly diplomatic was to get the authors ti fix it if it is just an ommission. BioC might be an alternative, because @jorainer 's Spectra package has the https://github.com/rformassspectrometry/MsBackendTimsTof. Yours, Steffen

[MatteoLacki]

Yes, I will try to get it there after 2 weeks. The license was on the github page, but, as mentioned, its omission was not pointed out by CRAN.

Best wishes,

Matteo

tel. +49 159 01681376

GitHub: MatteoLacki https://github.com/MatteoLacki

On Tue, Jun 13, 2023 at 3:18 PM Steffen Neumann @.***> wrote:

Thanks for the swift reply, yes, alternatively we could do install_github(), but something like CRAN or BioC still would make sense. Getting the LICENSE right is important, and there are certainly diplomatic was to get the authors ti fix it if it is just an ommission. BioC might be an alternative, because @jorainer https://github.com/jorainer 's Spectra package has the https://github.com/rformassspectrometry/MsBackendTimsTof. Yours, Steffen

— Reply to this email directly, view it on GitHub https://github.com/michalsta/opentims/issues/20#issuecomment-1589300296, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA6H2AFJ6IMQTUQ3YR4OL4LXLBSCFANCNFSM6AAAAAAZEQVMYI . You are receiving this because you commented.Message ID: @.***>

[michalsta]

Hi,

It's deeper than that actually, there was other reason CRAN people mentioned for removal: we use a builtin libzstddec quite deeply in our internals. As R has a package that does zstd, their policy says we should use that: pass back the data to R functions, then get them back to C++ code, then pass back to R internals. That would require us to implement a separate, R-specific code path in libopentims++, which will totally break the architecture of "opentimsr is just R API wrapper for C++ library libopentims++" for no technical reason at all other than "following policy". We tried appealing the removal, but the CRAN people didn't bother to dignify us even with a "no" response.

Seeing now that there are reverse dependencies and actual interest in the package I'll try to "fix" that.

[triston-groff]

Yes, I will try to get it there after 2 weeks. The license was on the github page, but, as mentioned, its omission was not pointed out by CRAN. Best wishes, Matteo tel. +49 159 01681376 GitHub: MatteoLacki https://github.com/MatteoLacki … On Tue, Jun 13, 2023 at 3:18 PM Steffen Neumann @.> wrote: Thanks for the swift reply, yes, alternatively we could do install_github(), but something like CRAN or BioC still would make sense. Getting the LICENSE right is important, and there are certainly diplomatic was to get the authors ti fix it if it is just an ommission. BioC might be an alternative, because @jorainer https://github.com/jorainer 's Spectra package has the https://github.com/rformassspectrometry/MsBackendTimsTof. Yours, Steffen — Reply to this email directly, view it on GitHub <#20 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA6H2AFJ6IMQTUQ3YR4OL4LXLBSCFANCNFSM6AAAAAAZEQVMYI . You are receiving this because you commented.Message ID: @.>

Hi,

It's deeper than that actually, there was other reason CRAN people mentioned for removal: we use a builtin libzstddec quite deeply in our internals. As R has a package that does zstd, their policy says we should use that: pass back the data to R functions, then get them back to C++ code, then pass back to R internals. That would require us to implement a separate, R-specific code path in libopentims++, which will totally break the architecture of "opentimsr is just R API wrapper for C++ library libopentims++" for no technical reason at all other than "following policy". We tried appealing the removal, but the CRAN people didn't bother to dignify us even with a "no" response.

Seeing now that there are reverse dependencies and actual interest in the package I'll try to "fix" that.

Hi,

I am wondering if there are any updates on this issue, or an expected timeline?

I tried installing opentimsr from source and devtools, but keep running into compatibility issues with Rtools and build versions of dependencies. Having an install option from a repository like BioC would be extremely useful if CRAN is continuing to be a pain.

I know this has been inconvenient for your team too - I appreciate your efforts!!

Take care, Triston