Re: The Inbox: Kernel-dtl.1683.mcz

Christoph Thiede via Squeak-dev <[email protected]>
Newsgroups gmane.comp.lang.smalltalk.squeak.general
Message-ID <a96c423c-0fed-4ccd-8f8a-296d91ff71d3@MX2018-DAG1.hpi.uni-potsdam.de>
Hi all,

I've merged the BrokenPromise compatibility via 61Deprecated-ct.26 et al. today and added a draft note to the release notes. https://github.com/hpi-swa/Squot/pull/412 is ready for Jakob's review to make Squot work properly with the compatibility version of BrokenPromise in Squeak 6.1. Any other packages that only handle BrokenPromise without assuming it's an error should work fully transparently in 6.1, though. :-)

So ... as soon as Jakob agrees with this PR, the showstopper should be no more. :-)

Best,
Christoph

-- 
Sent from Squeak Inbox Talk

On 2026-05-04T23:44:11+00:00, [email protected] wrote:

> Hi all,
> 
> I only had today a chance to actually test this change set with Squot. :o And it's not yet really working. I noticeed Promise>>error was also removed without deprecation, turning harmless get accesses into exceptions ... This needs to be deprecated properly. Also, Squit relies on the only invariant that my compat patch cannot restore, which is that BrokenPromise is no longer a subclass of Error. A fix for that on the Squot side is straightforward, though. I will make both changes tomorrow or the day after, I hope. Sorry I'm not having more time right now!
> 
> Independently of whether this is technically working, I'd still like to revise the decision of deprecating BrokenPromise in the future. Maybe we can describe this change in the release note as "not stable" or something?
> 
> > As a follow up, I would like to think of some way to remind ourselves to
> > update the Promise>>signalErrorValue method a year from now when the
> > deprecated BrokenPromise class is finally removed. Hmmm... maybe a
> > Promise that replies to this email after (Duration days: 365)
> > asMilliSeconds ;-)
> 
> I don't think we should ever make this cleanup, not even in Squeak 10.0, because at that point we would be breaking backward compatibility for pre 6.1. The good news is we also don't have to make that cleanup because it the patch is designed to not change the semantics of the remaining #signalErrorValue implementation unless someone uses the deprecated BrokenPromise class, though. :-)
> 
> Best,
> Christoph
> 
> Von: [email protected] <[email protected]>
> Gesendet: Dienstag, 5. Mai 2026 00:53 Uhr
> An: The general-purpose Squeak developers list <[email protected]>
> Betreff: [squeak-dev] Re: The Inbox: Kernel-dtl.1683.mcz 
>  
> On 2026-05-04 16:12, tim Rowledge wrote:
> 
> >> On May 4, 2026, at 7:23 AM, Marcel Taeumel (H) via Squeak-dev 
> >> <[email protected]> wrote:
> >> 
> >> Hi Dave, hi Christoph, hi Jacob --
> >> Christoph's changeset seems to work. Can we merge it?
> > 
> > Having just recently started using GitBrowser to try to get some 
> > Seaside fixes & improvements accepted, this would be a Good Thing.
> > 
> > tim
> > 
> 
> Yes please do merge it. Although I did not test it, I read the change 
> set and it all looks good to me.
> 
> As a follow up, I would like to think of some way to remind ourselves to 
> update the Promise>>signalErrorValue method a year from now when the 
> deprecated BrokenPromise class is finally removed. Hmmm... maybe a 
> Promise that replies to this email after (Duration days: 365) 
> asMilliSeconds ;-)
> 
> Thanks Christoph for providing the fix!
> 
> Dave
> Squeak-dev mailing list -- [email protected]
> To unsubscribe send an email to [email protected]

Squeak-dev mailing list -- [email protected]
To unsubscribe send an email to [email protected]
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.