Re: [VOTE] let construct (Block Scoping)

On Thu, Jan 22, 2026, 12:12 Rowan Tommins [IMSoP] <imsop.php@rwec.co.uk> wrote: > On 22/01/2026 09:12, Tim Düsterhus wrote: > > Hi > > > > we just opened the vote on the “let construct (Block Scoping)” RFC > > (which was originally proposed as “use construct”, but the keyword > > changed as a result of the discussion). > > > I have voted No. > > I really want block scoping as a feature in PHP, but I do not want this > syntax. > > > For those who didn't follow the discussion thread, I will try to > summarise my reasons: > > - I don't think PHP is special enough to break with a strong tradition > here. A huge family of languages use roughly the same syntax for > variable declarations, even though they have very different details of > how they work. That includes C and Java, but also JS and Perl, and even > VisualBasic (all ultimately traceable to ALGOL). > > - Users are likely to be coming from those related languages, > particularly JS, and are likely to be confused by how this syntax works. > > - For instance, they would not expect a comma-separated list to be > equivalent to a set of nested blocks, which affects the behaviour of > cases like let($a=new Foo, $a=new Bar) > > - The use cases for combining with "if" or "foreach" are interesting, > but the result is confusing - the important control flow keyword ends up > in the middle of the line. > > - For other use cases, requiring an extra block adds noise. For > instance, if we ever add auto-capture closures, a JS-style a concise > block-scoped declaration would be very useful to avoid accidental capture.. > > > PHP itself already has "ALGOL-style" declarations, for "const", > "global", and "static". I think adding a "let" or "var" > keyword with the > same syntax would be more consistent, and more useful. > > There would be details to work out, but dozens of other languages to > learn from. > > > My thanks to Seifeddine and Tim for their work on this, and to Tim in > particular for patient and respectful discussion. > > > -- > Rowan Tommins > [IMSoP] > I agree, the proposed syntax and nested block hell are terrible! The traditional syntax with implicit and invisible nested block are much better, many languages and years of trusting proved that. >