Updates to upstream asio 1.28.2

[shikokuchuo]

Closes #10.

This is the PR we've prepared @eddelbuettel.

On top of the update, there is one small patch, which ensures that snprintf is always used rather than sprintf, which R CMD check would also complain about.

Thanks!

[eddelbuettel]

I made some comments about things I think we should consider changing. Nothing super-set in stone but established best practice.

[eddelbuettel]

Well this is wrong because the commit is yours not mine.

Can we simplify and make it more truthful, ie you add a Changelog entry about the changes to inst/include/* and if possibly make no other changes?

A patch for sprintf -> snprint is sensible too, it may make sense to keep it in a local file (say local/change_to_1.28.2.patch or something like that,

[shikokuchuo]

I've limited this now to the library update and the patch.

For the patch, I think having the mention in the Changelog should be enough. It doesn't look like we update this library very often and down the line it may not even be needed - as long we run revdepchecks, we'd catch any future omission.

I'm not seeing a great benefit to keeping it in another file as we'd still need to patch asio.hpp to include it I suppose.

[eddelbuettel]

I still like to see the patch simply as a reminder.

And I don't mean in another file to be included. Simply as a 'note' of changes made due to CRAN. Look at my BH repo, I keep them there too (also to reapply them, of course, but it is more "industrial sized" there).

[shikokuchuo]

Right, got it. Done in a22fb6c.

[eddelbuettel]

'Files changed: 678' is very scary. Hard to review, but I trust you on the overall change to asio itself. I made some genty nudging proposals for other other minor details, let me know how you feel about those.

[shikokuchuo]

Thanks @eddelbuettel an impressively fast response! I've applied your suggestions in 061549c.

The large number of files changed are just the library, a straight inclusion from https://sourceforge.net/projects/asio/files/asio/1.28.2%20%28Stable%29/ (for reference).

[eddelbuettel]

Well the timing and timezones work in our favour. I was actually at the track for not quite an hour earlier, and was just gone for 2+ hours. Back now, all moving in the right direction and I can probably take care of 'shipping this' today.

[eddelbuettel]

FYI BH has the patches here: https://github.com/eddelbuettel/bh/tree/master/local/patches

[shikokuchuo]

Still impressive given all the repos you maintain! Thanks again.

[eddelbuettel]

Thank you -- we could add the patch file here, but I can also sneak that in when I do my iteration.

[eddelbuettel]

This is very clean now. Appreciate the changes.

[eddelbuettel]

@shikokuchuo One possible concern: Did you see the changelog / new entry from my last release? Apparently I had to switch a few upstream things to (v)snprintf. But besides the one you fixed the three reverse depends you looked did not toss up any sprintf issues? (I ran a quick reverse depends check too, but that is mostly for R CMD check fail and less about NOTEs.)

[eddelbuettel]

Actually the patch file answers that. I suppose upstream did not cast a net wide enough, so you got into trouble on macOS and this addresses this. Did you send that upstream too, just in case?

[shikokuchuo]

@shikokuchuo One possible concern: Did you see the changelog / new entry from my last release? Apparently I had to switch a few upstream things to (v)snprintf. But besides the one you fixed the three reverse depends you looked did not toss up any sprintf issues? (I ran a quick reverse depends check too, but that is mostly for R CMD check fail and less about NOTEs.)

This is fine. I just checked through those edits. In 1.28.2, those places all have a ASIO_HAS_SNPRINTF conditional branch now, so we're good with my patch.

[shikokuchuo]

Actually the patch file answers that. I suppose upstream did not cast a net wide enough, so you got into trouble on macOS and this addresses this. Did you send that upstream too, just in case?

I think you're referring to 'allocator<void>' is deprecated. That's a separate issue, and in #9 there's a link to a comment from the upstream author saying that's a false positive. We've dealt with it separately in the same way as in #9.

[eddelbuettel]

No I was thinking of snprintf but I rushed; looking at your patch again explains why it may not be for upstream. You are essentially ensuring Simon's old macOS-13 setup has it set too. We should be good.

[eddelbuettel]

Shipped. Bit of a queue there which needs to clear first but might be up in a few hours:

edd@rob:~/git/asioheaders(master)$ ciw.r                        # littler wrapper for ciw::ciw()
     Folder                          Name                Time   Size          Age
     <char>                        <char>              <POSc> <char>   <difftime>
 1: pretest   AsioHeaders_1.28.2-1.tar.gz 2025-04-08 19:09:00   677K   0.07 hours
 2: pretest            tickr_1.0.1.tar.gz 2025-04-08 18:47:00    49K   0.43 hours
 3: pretest             VLF_1.1-03.tar.gz 2025-04-08 18:37:00   496K   0.60 hours
 4: pretest             tcpl_3.2.1.tar.gz 2025-04-08 18:35:00   3.2M   0.63 hours
 5: pretest     PathwaySpace_1.0.1.tar.gz 2025-04-08 18:33:00   4.5M   0.67 hours
 6: pretest             puff_0.1.0.tar.gz 2025-04-08 18:33:00   298K   0.67 hours
 7: pretest    spanishoddata_0.1.1.tar.gz 2025-04-08 18:33:00   1.5M   0.67 hours
 8: pretest             solaR_0.47.tar.gz 2025-04-08 18:20:00   118K   0.88 hours
 9: pretest      rscorecard_0.31.0.tar.gz 2025-04-08 18:19:00   164K   0.90 hours
10: pretest     CloneSeeker_1.0.13.tar.gz 2025-04-08 18:11:00   301K   1.03 hours
11: pretest        featForge_0.1.1.tar.gz 2025-04-08 18:10:00    49K   1.05 hours
12: pretest SPARRAfairness_0.1.0.0.tar.gz 2025-04-08 17:56:00   1.0M   1.28 hours
13: pretest OptHoldoutSize_0.1.0.1.tar.gz 2025-04-08 17:42:00   2.9M   1.52 hours
14: pretest             heck_0.1.4.tar.gz 2025-04-08 17:40:00   567K   1.55 hours
15: pretest       crosswalkr_0.3.0.tar.gz 2025-04-08 17:36:00    72K   1.62 hours
16: pretest       treesitter_0.2.0.tar.gz 2025-04-08 17:30:00   381K   1.72 hours
17: pretest          popdemo_1.3-2.tar.gz 2025-04-08 17:29:00   3.4M   1.73 hours
18: pretest          DeSciDe_1.0.0.tar.gz 2025-04-08 17:28:00   868K   1.75 hours
19: pretest             beam_2.0.4.tar.gz 2025-04-08 17:24:00   260K   1.82 hours
20: pretest           artma_0.1.12.tar.gz 2025-04-08 17:20:00    37K   1.88 hours
21: pretest          FlexReg_1.3.1.tar.gz 2025-04-08 17:19:00   303K   1.90 hours
22: recheck         Rfast2_0.1.5.3.tar.gz 2025-04-08 17:17:00   169K   1.93 hours
23: recheck      formatters_0.5.11.tar.gz 2025-04-08 17:10:00   2.4M   2.05 hours
24: pending    GrowthCurveME_0.1.1.tar.gz 2025-04-08 17:09:00   412K   2.07 hours
25: recheck        rmumps_5.2.1-34.tar.gz 2025-04-08 16:08:00   2.2M   3.08 hours
26: recheck         cmsafops_1.4.1.tar.gz 2025-04-08 14:36:00   296K   4.62 hours
27: recheck          ggplot2_3.5.2.tar.gz 2025-04-08 12:59:00   3.3M   6.23 hours
28: waiting         pbkrtest_0.5.4.tar.gz 2025-04-02 17:30:00    76K 145.72 hours
     Folder                          Name                Time   Size          Age
edd@rob:~/git/asioheaders(master)$ 

[shikokuchuo]

Amazing! We've enough time on our side 😄

[eddelbuettel]

All done, not immediately as reckoned, but done:

Thanks for the high-quality PR, and I hope it helps the three posit packages dependencing on it.