On 2017/12/02 01:59, Klemens Nanni wrote:
> On Fri, Dec 01, 2017 at 02:26:06PM +0000, Stuart Henderson wrote:
> > : COMMENT = Faster Go re-implementation of sloccount for more
> > languages
> >
> > just describe what it does rather than say things like "faster
> > reimplementation" and "more". in particular it's quite useless to
> > have a package comparing itself with something else that isn't in
> > packages.
> Fair points, I've updated COMMENTS and simply included SLOCCount's
> description from https://www.dwheeler.com/sloccount/ (mostly matching
> what benno@ used) in DESCR.
>
> > : WANTLIB = c pthread
> >
> > loccount-1.1(textproc/loccount):
> > Extra: c.92 pthread.25
> >
> > if you're adding WANTLIB in the hope that it triggers an update in a
> > statically
> > linked program, add a comment explaining why, otherwise it is likely to be
> > removed
> > in a WANTLIB sync.
> Will "# statically linked" suffice?
Yes.
> > : MODULES = lang/go
> >
> > there doesn't seem much advantage to using the lang/go module when
> > you're not using the go build scaffolding.
> >
> > : ALL_TARGET = ${HOMEPAGE:https://%=%}
> >
> > this is a bit obtuse, if you were actually using it, it would be simpler
> > to just use the string.
> >
> > but you're not actually using it.
> Yeah, I started with MODULES=lang/go basically and made it "work"...
>
> > : WRKSRC = ${MODGO_WORKSPACE}/src/${ALL_TARGET}
> > :
> > : do-build:
> > : make -C ${WRKSRC} loccount{,.1}
> > :
> > : do-test:
> > : make -C ${WRKSRC} check
> >
> > normally ${MAKE_PROGRAM} rather than calling make directly, but seems
> > simpler to get rid of the MODULES and have something like this instead
> >
> > BUILD_DEPENDS = textproc/asciidoc \
> > lang/go
> >
> > ALL_TARGET = loccount loccount.1
> >
> > do-install:
> > ${INSTALL_PROGRAM} ${WRKSRC}/loccount-$V-* ${PREFIX}/bin/loccount
> > ${INSTALL_MAN} ${WRKSRC}/loccount.1 ${PREFIX}/man/man1/
> >
> ... or not. This port is made around it's Makefile not the Go build
> environment as seen in my MODULES approach; I went through this multiple
> times now and ditching it in favor of BUILD_DEPENDS is indeed clearer.
>
> How about this diff on top of my tarball?
Please send updated tarballs for changes to a new port - a diff in the
accompanying mail can help to explain changes, but it's better not to
ask reviewers to dig through old mails to assemble the whole thing.
> The symlink in post-build makes the test work and avoids -${V}-* in
> install.
>
> Thanks a lot for your feedback!
>
> ---
> textproc/loccount/Makefile | 19 ++++++++-----------
> textproc/loccount/pkg/DESCR | 19 +++++++++++++++++++
> 2 files changed, 27 insertions(+), 11 deletions(-)
>
> diff --git a/textproc/loccount/Makefile b/textproc/loccount/Makefile
> index 8a8319e83eb..a263287844a 100644
> --- a/textproc/loccount/Makefile
> +++ b/textproc/loccount/Makefile
> @@ -1,6 +1,6 @@
> # $OpenBSD: $
>
> -COMMENT = Faster Go re-implementation of sloccount for more
> languages
> +COMMENT = Count lines of codes in many languages
lowercase Count
> V = 1.1
> DISTNAME = loccount-${V}
> @@ -16,20 +16,17 @@ MASTER_SITES = ${HOMEPAGE}/repository/
> # BSD
> PERMIT_PACKAGE_CDROM = Yes
>
> +# statically linked
> WANTLIB = c pthread
>
> -MODULES = lang/go
> +BUILD_DEPENDS = lang/go \
> + textproc/asciidoc
>
> -BUILD_DEPENDS = textproc/asciidoc
> +ALL_TARGET = loccount{,.1}
not a fan of relying on shell expansions for a make variable unless
there's no other way, it's clearer like this:
ALL_TARGET = loccount loccount.1
> +TEST_TARGET = check
>
> -ALL_TARGET = ${HOMEPAGE:https://%=%}
> -WRKSRC = ${MODGO_WORKSPACE}/src/${ALL_TARGET}
> -
> -do-build:
> - make -C ${WRKSRC} loccount{,.1}
> -
> -do-test:
> - make -C ${WRKSRC} check
> +post-build:
> + ln -sf ${WRKBUILD}/loccount{-${V}-*,}
>
> do-install:
> ${INSTALL_PROGRAM} ${WRKSRC}/loccount ${PREFIX}/bin/
> diff --git a/textproc/loccount/pkg/DESCR b/textproc/loccount/pkg/DESCR
> index 798b459ddb7..8f58a059091 100644
> --- a/textproc/loccount/pkg/DESCR
> +++ b/textproc/loccount/pkg/DESCR
> @@ -6,3 +6,22 @@ implementation of the original.
> The algorithms are largely unchanged and can be expected to produce identical
> numbers for languages supported by both tools. Python is an exception;
> loccount
> corrects buggy counting of single-quote multiline literals in sloccount 2.26.
> +
> +
> +SLOCCount includes a number of heuristics, so it can automatically detect
> file
> +types, even those that don't use the "standard" extensions, and conversely,
> it
> +can detect many files that have a standard extension but aren't really of
> that
> +type. The SLOC counters have enough smarts to handle oddities of several
> +languages. For example, SLOCCount examines assembly language files,
> determines
> +the comment scheme, and then correctly counts the lines automatically. It
> also
> +correctly handles language constructs that are often mishandled by other
> tools,
> +such as Python's constant strings when used as comments and Perl's "perlpod"
> +documentation.
> +
> +SLOCCount will even automatically estimate the effort, time, and money it
> would
> +take to develop the software (if it was developed as traditional proprietary
> +software). Without options, it will use the basic COCOMO model, which makes
> +these estimates solely from the count of lines of code. You can get better
> +estimates if you have more information about the project; see the SLOCCount
> +documentation for information on how to control the estimation formulas used
> in
> +SLOCCount.
>
This still makes the whole file revolve around the fact that it's a
reimplementation of sloccount. That's really of minor interest to DESCR,
the primary function is to say what the package does - if the details of
a similar program are needed at all then one paragraph at the bottom
should suffice?
(Something like "loccount is a re-implementation of David A. Wheeler's
sloccount tool in Go. The algorithms are largely unchanged and, bugs
excepted, can be expected to produce identical numbers for languages
supported by both tools." perhaps ..)
