Re: PORTSCOUT= position in port Makefile
Daniel Engberg <[email protected]> Sat, 6 Jun 2026 08:55:05 +0200
| Newsgroups | gmane.os.freebsd.devel.ports |
|---|---|
| Message-ID | <[email protected]> |
On 2026-06-06 08:33, Don Lewis wrote: > Two chapters of the FreeBSD Porter's Handbook discuss the ordering of > the variables in the port Makefile: > Chapter 5. Configuring the Makefile > https://docs.freebsd.org/en/books/porters-handbook/makefiles/ > > Chapter 15. Order of Variables in Port Makefiles > https://docs.freebsd.org/en/books/porters-handbook/order/ > > PORTSCOUT is mentioned in Section 5.9: > https://docs.freebsd.org/en/books/porters-handbook/makefiles/#makefile-portscout > where is is shown following the LICENSE block. > > PORTSCOUT is not mentioned explicitly in Chapter 15, in which case > 15.11. The Rest of the Variables, should apply: > "And then, the rest of the variables that are not mentioned in the > previous blocks, placing PORTSCOUT right before the Targets." > > Portlint and portclippy both seem to prefer the Chapter 15 ordering. > > Neither position really makes sense to me. Since PORTSCOUT relates to > the PORTVERSION, it makes the most sense to me to include it in the > PORTNAME block. > > Thoughts? Neither of these tools are bulletproof and not expected to be. Portscout isn't related to the port metadata so I'd argue that below LICENSE block is fine and it makes less sense to traverse though a complete (port) Makefile to find it mixed in with build related variables. Below LICENSE block is also how most Makefiles used to define it and I think there's simply an oversight in Chapter 15. While one the topic, Portscout itself is falling more or less into obsolescence since it can't cope with various developer platforms and other "quirks". Perhaps add on the long list of many "todo"/"things that needs to be resolved" regarding the Ports tree list add move to nvchecker or similar tool which likely will use separate file See https://gitlab.archlinux.org/archlinux/packaging/packages/curl/-/blob/main/.nvchecker.toml?ref_type=heads for example Best regards, Daniel