Check for Updates resolves 25.2.6 available version, but when I click on "Download" the webpage shows 25.8.2. Why?

This exact same thing has happened before (with older versions, of course).

When I click on “Check for Updates”, it shows some version (25.2.6 in this case), and says that I have to go and download installer. And when I click on Download, it takes me to the download page where the latest versions is 25.8.2.

image

Why this update system is always working so inconsistently?

Could you please also provide screenshot of the lower part of the page?

And, again - this inconsistency of versions suggested in the Check for Update window and on the website - is very prevalent, as it happens often.

Ah… look at this: https://ask.libreoffice.org/t/version-7-0-3-announced-while-version-7-0-4-available/

So:

  1. The version 25.8 is not tested long enough, to be recommended for replacement of old stable releases (it is only in .2 bugfix release stage). Only when it reaches 25.8.4 (or .5), will it be recommended. Therefore, more tested 25.2 is recommended.
  2. In every branch, the autoupdater starts recommending the newest release with a lag of ~2 weeks. Therefore, it’s still not informed about 25.2.7, and recommends 25.2.6.

Aha, and @cloph even answered your old question himself :slight_smile:

Wow… that is pretty bad.

First of all, if 25.8.2 is “not tested long enough”, then why it’s advertised as “latest stable release”? This is quite contradictory and confusing. Very much so;

Secondly, the auto-updater did not recommend “the newest release”, as you say, it has recommended the release which is not even found on the website’s download page;

Finally - if that ~2 weeks lag is present, why it’s not communicated either on the page or in the auto update window? It’s really really confusing and obscure. And even if it’s a lag of ~2 weeks, still - it should be clear to WHICH EXACT release this lag is attributed to. There’s no way an average user can automagically guess that “25.2 is more recommended”, as you say. I, as an user, should have a clear picture of how to update my libreoffice, without being a core engine branch hacker of LO.

Yes, he/she has answered, that: “It was a simple typo” - so, a mistake on development team side. But this is not something I should be seeing per every update. And again… what is the problem now? a typo? branching strategy? unclarity on the page and in the update window?.. consistency, please, and clarity. :slight_smile:

It is there on the page - at the very bottom.
But well, I don’t think that this UX problem that you describe is considered serious (but maybe @cloph and @ilmari would correct me). Personally I myself would agree with you that it is not perfect right now.

Mate, let’s appreciate KISS. System offers update, you go and click update. Period. End of story. It should never trigger such a long hackerrish deep low-level branching strategy discussions or etc. If the software architect gets confused and lost, I can imagine what would happen with a “normal” average user. And the funny thing is, that even in that discussion - arguments don’t make sense, cause this is bad design.

What is your point here writing that to me?

What? What is your point of writing anything here at all? Same for me - expanding and elaborating on this problem. Nothing more.

My point was: to provide facts. And also - to ping some people that could provide more facts, and/or consider your feedback (and I even hinted, that I do not consider the state ideal; those are just facts, not something I try to convince you to be how it should). But your “Mate, let’s appreciate KISS” is like arguing with me - in what?

I can’t imagine what more facts you need, I provided all. And I don’t know who are “some people” I should ping.

Regarding arguing, it’s now you who’s arguing - not me. I just expressed what it should be like.

Also, so what should be the “source of truth” trust point - version suggested in the auto-updated window or the version on the website? And how would I know if the “latest stable” advertised version is not really tested and latest stable one?

Lol. I tried to provide facts, that was my point, sigh…

Ok. you provided facts and you have brought up this topic. End of discussion.

And the latest “well tested” solution is maintained until November. Amazing.

But really - what should I rely on? Auto-updater window or the website which claims “latest stable” is the release which you say is not tested well? Which one should I trust usually?

Giorgi: you have made many statements in this thread. What is your main point? That the strategy of having two parallel branches should be abandoned?

That updating LibreOffice is very confusing, unclear and misleading, saying one thing in the app and completely different thing on the website. That it doesn’t provide any information mentioned in this thread (that “latest stable” release is apparently “not well tested”). That I want to know what update strategy I should follow when updating LO.

Are any of these statements not worth attention? Maybe you could thank me for providing feedback, instead?

The 25.2.6 vs. 25.2.7 discrepancy is something our release engineer, who was already pinged in this thread, might answer.

The general note of something being “not well tested” is just referring to the count of the minor versions released for the particular major version. The strategy is to not add features in minor releases and thus minimising the potential for breakage. Each minor release hopefully fixes more bugs. Even the first release of a major version is considered stable. An unstable release is an unreleased one, such as the nightly builds we offer for testers.