Re: tlcritical installed tlmgr from TL26 in TL25
Norbert Preining <[email protected]>
| Newsgroups | gmane.comp.tex.live |
|---|---|
| Message-ID | <aYpny1h1fgAmQusF@burischnitzel> |
Hi Karl, hi Max, > Yikes. Clearly not expected, or desirable. I never realized that was the > behavior. I guess no one else has ever had tlcritical enabled by > default in all these years. I don't myself. I **do** but I have never seen this. > Existing practice is/has been the opposite of what Norbert said. (Can (sorry!!) > that be right?! I *think* so. :) There is only one tlcritical repo, and > when the pretest rolls around, I update it for the pretest, because > that's what's changing. Put another way, the assumption is that tlnet > will not have any critical changes once the pretest starts. So far, > that assumption has proven true. Hmmm, my brain is getting weak, but I thought it is the other way round, that tlcritical is built from the branch that is going into the normal tlnet. This might be a regression from the time we changed to update normal tlnet during pretest. Before that, we stopped updating tlnet completely during pretest, and probably also tlcritical. > I'm not entirely sure how to defend against updating the > tlcritical-for-pretest in tlnet. I don't think there's an easy way for > tlmgr to know what version is in tlcritical before installing any As I said, I think tlcritical should be built from the branch that targets tlnet. And tlpretest is AFAIR already updating the telxive-infra package on each rebuild? As said, I might be completely wrong, but this might have been introduced and not detected when we switched to update tlnet during tlpretest builds. Best regards Norbert -- DI Dr Norbert Preining https://www.preining.info arXiv / Cornell University + IFMGA Guide + TU Wien + TeX Live GPG: 0x860CDC13 fp: F7D8 A928 26E3 16A1 9FA0 ACF0 6CAC A448 860C DC13