summaryrefslogtreecommitdiff
path: root/pkgtools
AgeCommit message (Collapse)AuthorFilesLines
2006-05-31Added a check for variables that are evaluated at load time. Especiallyrillig2-13/+58
for lists of something, this can lead to unexpected behavior. Currently, only the variables BUILDLINK_DEPENDS, BUILDLINK_PACKAGES and PKG_OPTIONS may be looked at during load time. This warning reveals the bad practice to "patch" CONFIGURE_ARGS at load time, for example: > CONFIGURE_ARGS:= ${CONFIGURE_ARGS:S/--disable-esd/--enable-esd/} WARN: audio/bmp-esound/Makefile:14: CONFIGURE_ARGS should not be evaluated at load time.
2006-05-31Fixed indentation level in checkline_mk_vartype.rillig1-29/+29
2006-05-31The checks for variable use and correct quoting have been concentratedrillig2-107/+298
into a single subroutine, checkline_mk_varuse, which replaces the various ad-hoc checks. Added the class PkgLint::VarUseContext that tries to model the context in which a make(1) variable can appear. The checks are much better now than before, but there's still work to do. Added a new type FileMode for file permissions. Currently no warnings are printed for untyped variables (that is, user-defined and not following the common naming schemes). This includes the iteration variables of .for loops. Since many of the warnings have been overly strict, this is not a big loss.
2006-05-31Some minor fixes and additions.rillig1-3/+5
2006-05-31SHLIB_HANDLING and USE_RMAN are deprecated.rillig1-1/+3
2006-05-28Remove EOL comments for packages which were uncommented; theywiz1-3/+3
confused make.
2006-05-26After some discussion with jlam and no objection raised by agc,joerg1-4/+4
uncomment libnbcompat, mtree and xpkgwedge.
2006-05-23Moved the code that guesses the type of a variable based on its namerillig1-36/+40
into get_variable_type(). Added a new field PkgLint::Type::is_guessed that can be checked later.
2006-05-23Prevented warnings when defining the new FETCH_MESSAGE.rillig2-2/+5
2006-05-23- _FETCH_MESSAGE, BUILDLINK_RECOMMENDED.* and BUILDLINK_DEPENDS.* arerillig3-11/+26
deprecated. - CONFLICT is just a spelling mistake and therefore should not appear in the list of deprecated variables. - Made handling of parameterized variables more uniform.
2006-05-23The "future directions" chapter gives a vision of what pkglint may do ifrillig3-2/+53
I find too much time.
2006-05-22Move the check-shlibs target from bsd.pkg.mk to bsd.pkg.check.mk wherejlam1-2/+3
it will live with other "check" targets run after package installation. Get rid of SHLIB_HANDLING, whose meaning had mutated over the years from one thing to another. Currently, it is used to basically note whether the system's "ldd" command can be usefully run on the package's binaries and libraries. Rename this variable to CHECK_SHLIBS_SUPPORTED for more clarity. CHECK_SHLIBS is now a variable set exclusively by the user in /etc/mk.conf to note whether the check for missing run-time search paths is performed after a package is installed. It defaults to "no" unless PKG_DEVELOPER is set.
2006-05-22- When building the list of used variables, scan Makefile.* in additionrillig1-2/+2
to *.mk.
2006-05-22- Pkglint distinguishes between variables FOO and FOO.* now when checkingrillig2-202/+282
data types and permissions. - Changed the way that permissions are determined a bit, so that unspecified permissions can be detected. - Renamed -Wacl to -Wperm. - When no applicable permissions are found, "?" is returned. - When checking for unused variables, parameterized variables don't have to be matched exactly but only by basename. - Added an explanation for the permissions warning.
2006-05-22Removed some drivel that was not appropriate for man pages.rillig2-15/+3
2006-05-21- PLISTs should not contain perllocal.pod.rillig1-3/+9
- Variables called *_HOME are assumed to contain a single pathname.
2006-05-20Switching pkglint to OCaml looks feasible.rillig1-1/+10
2006-05-20Switching the programming language of pkglint to C++ is not an option.rillig1-1/+5
2006-05-19Reindented the whole program to fit my personal taste.rillig1-1395/+1441
2006-05-19- Found three more instances where explain() had been called instead ofrillig1-14/+33
explain_warning(), which aborted the program. - Added support for a wip-specific TODO file where suggested package updates are recorded.
2006-05-16Updated pkglint to 4.62.rillig2-4/+7
Changes since 4.61: - Bug fixes.
2006-05-16bsd.options.mk may set PKG_OPTIONS.rillig2-4/+6
2006-05-16Fixed a typo: $op cannot be "+", but sometimes is "+=".rillig1-2/+2
2006-05-15Found two instances of explain() that had not been converted properly.rillig1-4/+12
Fixed them. Added explain_note().
2006-05-15Allow the compressed form of duplicate elimination for BUILDLINK_PACKAGES.rillig1-9/+22
2006-05-13Updated pkglint to 4.61.rillig1-2/+2
Changes since 4.60: - Added ACL-based checks that restrict the files which may define variables. - Many, many small improvements.
2006-05-13Changed two lines in the generated buildlink3.mk files. For quite a longrillig2-6/+6
time, the !empty(BUILDLINK_DEPTH:M+) lines confused me, so I changed them to be ${BUILDLINK_DEPTH} == "+", which has the same effect. Changed the version number to 3.12.
2006-05-13- Added the constant NO_LINES as an alias for NO_LINE_NUMBER.rillig1-48/+84
- Renamed explain() to explain_warning() and added explain_info(). - Added some code to distinguish FOO from FOO.* in the makevars.map file. - Allow an alternative form of checking the inclusion depth in buildlink3.mk files.
2006-05-11In the make targets show-* and *-message, commands may be hidden usingrillig2-23/+30
"@".
2006-05-11Removed the "Comment expected" warning from patch files, as suggested byrillig1-5/+5
many people on pkgsrcCon 2006.
2006-05-11BUILDLINK_A{B,P}I_DEPENDS should not be assigned using the ?= operator.rillig1-3/+3
2006-05-11- Allow the lines in makevars.map to be folded like in Makefiles.rillig2-9/+15
- Removed some legacy definitions for ACL subjects.
2006-05-11Some ideas about the contexts in which variables may be used.rillig1-1/+40
2006-05-10- ACL subjects may contain * as a placeholder.rillig3-24/+30
2006-05-10The := or != operators can be used for unconditional assignment, too.rillig1-2/+3
2006-05-10clean up after doing a quick-install.rillig1-2/+2
2006-05-10- Improved ACL definitions in makevars.map.rillig3-72/+94
- Added checks for deprecated and nonexistent licenses. (suggested by gdt) - Added detection for /foo/s/bar/baz/ sed command that appears without a leading "-e" in a SedCommands variable. (suggested by wiz)
2006-05-10ACL entries are separated by a comma and optional white-space, since thatrillig1-3/+3
looks nicer.
2006-05-10- Removed the types Readonly and Userdefined.rillig2-239/+299
- Aligned the ACL permission letters with the documentation. The old "c" letter has been removed because from the definition's point of view, there is no difference between the "=" and the ":=" operator. All the differences lie on the right hand side, which will be covered later. - Updated the variables in makevars.map with mk/defaults/mk.conf, 1.118.
2006-05-10Rewrote the section on ACLs. Now I must adjust the implementation torillig1-61/+88
this specification.
2006-05-10* In the vartypes.map, distinguish FOO and FOO.*.rillig1-1/+2
2006-05-10- Added the class PkgLint::Type.rillig4-316/+461
- Added the -Wacl command line option, which is disabled by default. - Extended the type definitions in makevars.map by ACLs, which specify in which files the variables may be defined. - Trailing white-space in DESCR files can be fixed automatically. - Use of the $$ shell variable is not flagged as "strange" anymore. - Added ACL-based checks for variable definitions. - PKGREVISION should not be set to 0. - Improved the diagnostics text for RESTRICTED. - Added the types Unchecked and Version. - Dashes are allowed in library names in wrapper reordering commands. - More than one DISTFILES definition is allowed in the variable ordering check.
2006-05-10Make pkg_install build on platforms that don't define the PRIu64 integerjlam2-4/+6
format modifier. We include "defs.h" in the files that use PRIu64 as that is the header that contains stand-in definition.
2006-05-09Took maintainership.rillig1-2/+2
2006-05-06PKGMANDIR instead of "man".reed1-5/+5
2006-05-06This doesn't use c, c++, fortran, java, or objc.reed1-1/+2
2006-05-04Bump version to 1.72:dillo3-21/+22
Make it use pkg_summary(5) (with a fallback to the old pkg_chk-summary). Retire capability to create pkg_chk-summary.
2006-05-02In Makefiles, the level of indentation of directives should not berillig1-11/+16
affected by multiple-inclusion guards.
2006-05-02Non-optimal white-space is not worth a warning. A note is enough.rillig1-2/+2
2006-05-01It is an error if a RESTRICTED package does not have a LICENSE.rillig1-1/+5
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-30 19:24:48 +0000