February 2007 - macports-dev - lists.macosforge.org

portindex output depends on the locales
by Vincent Lefevre 16 Feb '07

16 Feb '07

Hi, The output of the portindex commands depends on whether the character set is UTF-8 or, e.g., ISO-8859-1, due to non-ASCII characters in a description: the number after the port name seems to contain the number of characters instead of the number of bytes. For instance, I get: stardict-xmlittre 310 in ISO-8859-1, but stardict-xmlittre 308 in UTF-8. Is this difference important? -- Vincent Lefèvre <vincent(a)vinc17.org> - Web: <http://www.vinc17.org/> 100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/> Work: CR INRIA - computer arithmetic / Arenaire project (LIP, ENS-Lyon)

1 1

Re: [22007] trunk/base/src/port1.0/portutil.tcl
by Jordan K. Hubbard 16 Feb '07

16 Feb '07

While most of these helper functions seem pretty sane, I can't help but think that the delete function really ought to be written in C for speed and correctness (following the pseudo code example at http://docs.info.apple.com/article.html?artnum=107884) The Tcl function I see below is not going to be particularly fast on either broad or deep hierarchies of files, and mass deletions of stuff in cleanup procedures is fairly common place. I think the other functions are low-overhead enough that Tcl is a fine implementation language, but not delete. - Jordan On Feb 13, 2007, at 7:09 PM, source_changes(a)macosforge.org wrote: > Revision > 22007 > Author > eridius(a)macports.org > Date > 2007-02-13 19:09:46 -0800 (Tue, 13 Feb 2007) > Log Message > > Reimplement delete so it no longer uses system "/bin/rm -rf ..." > This also fixes the case of filenames with spaces in them being > incorrectly parsed - That was EXTREMELY dangerous > Add new touch command - usage similar to BSD touch > Add new copy and move commands - basically aliases for file copy/ > rename > Add new ln command - usage similar to BSD ln > These should be documented somewhere > Modified Paths > > trunk/base/src/port1.0/portutil.tcl > Diff > > Modified: trunk/base/src/port1.0/portutil.tcl (22006 => 22007) > > --- trunk/base/src/port1.0/portutil.tcl 2007-02-13 22:54:21 UTC > (rev 22006) > +++ trunk/base/src/port1.0/portutil.tcl 2007-02-14 03:09:46 UTC > (rev 22007) > @@ -443,7 +443,7 @@ > proc ldelete {list value} { > set ix [lsearch -exact $list $value] > if {$ix >= 0} { > - return [lreplace $list $ix $ix] > + return [lreplace $list $ix $ix] > } > return $list > } > @@ -511,14 +511,204 @@ > } > > # delete > -# fast (and more reliable than 'file delete') file- and directory- > remove routine > +# file delete -force by itself doesn't handle directories properly > +# on systems older than Tiger. However we can recurse this thing > ourselves > proc delete {args} { > - foreach arg $args { > - ui_debug "delete: $arg" > - system "/bin/rm -rf $arg" > - } > + foreach arg $args { > + ui_debug "delete: $arg" > + set stack [list $arg] > + while {[llength $stack] > 0} { > + set file [lindex $stack 0] > + if {[file isdirectory $file]} { > + # it's a directory > + set children [glob -nocomplain -directory $file > {{*,.*}}] > + set children [ldelete [ldelete $children $file/.] > $file/..] > + if {[llength $children] > 0} { > + set stack [concat $children $stack] > + } else { > + # directory is empty > + file delete -force -- $file > + set stack [lrange $stack 1 end] > + } > + } else { > + # it's not a directory - kill it now > + file delete -force -- $file > + set stack [lrange $stack 1 end] > + } > + } > + } > } > > +# touch > +# mimics the BSD touch command > +proc touch {args} { > + while {[string match -* [lindex $args 0]]} { > + set arg [string range [lindex $args 0] 1 end] > + set args [lrange $args 1 end] > + switch -- $arg { > + a - > + c - > + m {set options($arg) yes} > + r - > + t { > + set narg [lindex $args 0] > + set args [lrange $args 1 end] > + if {[string length $narg] == 0} { > + return -code error "touch: option requires an > argument -- $arg" > + } > + set options($arg) $narg > + set options(rt) $arg ;# later option overrides > earlier > + } > + - break > + default {return -code error "touch: illegal option -- > $arg"} > + } > + } > + > + # parse the r/t options > + if {[info exists options(rt)]} { > + if {[string equal $options(rt) r]} { > + # -r > + # get atime/mtime from the file > + if {[file exists $options(r)]} { > + set atime [file atime $options(r)] > + set mtime [file mtime $options(r)] > + } else { > + return -code error "touch: $options(r): No such > file or directory" > + } > + } else { > + # -t > + # parse the time specification > + # turn it into a CCyymmdd hhmmss > + set timespec {^(?:(\d\d)?(\d\d))?(\d\d)(\d\d)(\d\d)(\d > \d)(?:\.(\d\d))?$} > + if {[regexp $timespec $options(t) {} CC YY MM DD hh mm > SS]} { > + if {[string length $YY] == 0} { > + set year [clock format [clock seconds] -format > %Y] > + } elseif {[string length $CC] == 0} { > + if {$YY >= 69 && $YY <= 99} { > + set year 19$YY > + } else { > + set year 20$YY > + } > + } else { > + set year $CC$YY > + } > + if {[string length $SS] == 0} { > + set SS 00 > + } > + set atime [clock scan "$year$MM$DD $hh$mm$SS"] > + set mtime $atime > + } else { > + return -code error \ > + {touch: out of range or illegal time > specification: [[CC]YY]MMDDhhmm[.SS]} > + } > + } > + } else { > + set atime [clock seconds] > + set mtime [clock seconds] > + } > + > + # do we have any files to process? > + if {[llength $args] == 0} { > + # print usage > + ui_msg {usage: touch [-a] [-c] [-m] [-r file] [-t [[CC]YY] > MMDDhhmm[.SS]] file ...} > + return > + } > + > + foreach file $args { > + if {![file exists $file]} { > + if {[info exists options(c)]} { > + continue > + } else { > + close [open $file w] > + } > + } > + > + if {[info exists options(a)] || ![info exists options(m)]} { > + file atime $file $atime > + } > + if {[info exists options(m)] || ![info exists options(a)]} { > + file mtime $file $mtime > + } > + } > + return > +} > + > +# copy > +proc copy {args} { > + exec file copy $args > +} > + > +# move > +proc move {args} { > + exec file rename $args > +} > + > +# ln > +# Mimics the BSD ln implementation > +# ln [-f] [-h] [-s] [-v] source_file [target_file] > +# ln [-f] [-h] [-s] [-v] source_file ... target_dir > +proc ln {args} { > + while {[string match -* [lindex $args 0]]} { > + set arg [string range [lindex $args 0] 1 end] > + set args [lrange $args 1 end] > + switch -- $arg { > + f - > + h - > + s - > + v {set options($arg) yes} > + - break > + default {return -code error "ln: illegal option -- $arg"} > + } > + } > + > + if {[llength $args] == 0} { > + ui_msg {usage: ln [-f] [-h] [-s] [-v] source_file > [target_file]} > + ui_msg { ln [-f] [-h] [-s] [-v] file ... directory} > + return > + } elseif {[llength $args] == 1} { > + set files $args > + set target ./ > + } else { > + set files [lrange $args 0 [expr [llength $args] - 2]] > + set target [lindex $args end] > + } > + > + foreach file $files { > + if {[file isdirectory $file] && ![info exists options(s)]} { > + return -code error "ln: $file: Is a directory" > + } > + > + if {[file isdirectory $target] && ![info exists options > (h)]} { > + set linktarget [file join $target [file tail $file]] > + } else { > + set linktarget $target > + } > + > + if {[file exists $linktarget] && ![info exists options(f)]} { > + return -code error "ln: $linktarget: File exists" > + } > + > + if {[llength $files] > 2} { > + if {![file exists $linktarget]} { > + return -code error "ln: $linktarget: No such file > or directory" > + } elseif {![file isdirectory $target]} { > + # this error isn't striclty what BSD ln gives, but > I think it's more useful > + return -code error "ln: $target: Not a directory" > + } > + } > + > + if {[info exists options(v)]} { > + ui_msg "ln: $linktarget -> $file" > + } > + if {[info exists options(s)]} { > + file link -symbolic $linktarget $file > + } else { > + file link -hard $linktarget $file > + } > + } > + return > +} > + > # filefindbypath > # Provides searching of the standard path for included files > proc filefindbypath {fname} { > _______________________________________________ > macports-changes mailing list > macports-changes(a)lists.macosforge.org > http://lists.macosforge.org/mailman/listinfo/macports-changes

2 2

[PATCH] Postfix port
by Gufo 15 Feb '07

15 Feb '07

Hi all, this is my first mail here, please correct if is not the right place to post this kind of informations. The current port of postfix doesn't work well, so I did some work trying to make it ok: Attached are: - a simple patch to postfix-script so it can correctly recognise folders permissions - a patch to the portfile which: * add cyrus support (this must be done explicitely in 2.3.X) * try to compile postfix against _system_ openssl and _macports'_ sasl2 * bump postfix version to 2.3.4 There's one unresolved issue, no matter if I explicitly compile it against system or macports version of openssl, I have this warning: ------ postfix/smtp[24927]: warning: run-time library vs. compile-time header version mismatch: OpenSSL 0.9.8 may not be compatible with OpenSSL 0.9.7 ------ Which means it compiles against system version of openssl and uses macports' one at run time. As a result I can't start a tls connection with any mailserver. After installation 2 files must be deleted (I don't know which is the best way to do this in the Portfile): /opt/local/var//spool/postfix/maildrop/.turd_postfix /opt/local/var//spool/postfix/public/.turd_postfix Please let me know what do you think about this, and forgive me if something is wrong because this is my first attempt to modify a port. Gufo -- ,___, If you're not part of the solution, (9v9) you're part of the precipitate (_^((\ My site: http://gufo.dontexist.org/ -^-"-"-\\-^-- Gpg Key at: http://gufo.dontexist.org/db.asc

5 5

NEW: gtk2-extra
by Anant Narayanan 15 Feb '07

15 Feb '07

Hi, Small reminder for a pending new port I posted sometime ago: http://article.gmane.org/gmane.os.apple.macports.devel/285/ Thanks, -- Anant

2 1

How to submit a new port ?
by Regis Vincent 15 Feb '07

15 Feb '07

hi, I'm not sure this is the right mailing list (from the archive it seems so), I have made a new portfile for an open-source project called playerstage (see playerstage.sourceforge.net) but I don't how to submit this port to the main tree (macport !) ? I even offer to maintain this port for a while but there is no documentation on how to submit a new port. The two portfiles are here: http://www.ai.sri.com/~vincent/Port/science/player/Portfile for player http://www.ai.sri.com/~vincent/Port/science/stage/Portfile for stage R.

2 1

Re: [22007] portutil.tcl mods / more portfile.7 changes needed?
by Mark Duling 15 Feb '07

15 Feb '07

"Jordan K. Hubbard" <jkh(a)opendarwin.org> on Wednesday, February 14, 2007 at 12:31 AM -0800 wrote: > >xinstall and file copy are different - xinstall is implemented in >pextlib, and apparently doesn't recursively copy directories. It's a >shame. > > >So yeah, while xinstall won't handle directories, copy will do it just >fine. The difference is that copy won't handle file permissions, but >xinstall does. > > >I was trying to copy the semantics of install(1) in order to obey what >seemed to be the principle of least astonishment. That said, I'm having >a hard time thinking of just how and where the world would end if someone >were to enhance it to actually copy directories. I was probably being >too conservative. For the sake of consistency, it would be a great thing. From what I understand from Kevin's explanation last night, to copy recursively and set permissions without system calls requires this: ---------- copy ${worksrcpath}/src/lib ${destroot}${prefix}/lib/${name} find ${destroot}${prefix}/lib/${name} {expr 1} { file attributes $filename -permissions 755 -owner root -group wheel --------- Which really in't so bad, but it would be nicer if it could be done in a single commd. But for now should this be the recommended way? If so I think we need to stick this in portfile.7. Mark

1 0

Re: Anyone watching the bugtracker ?
by Juan Manuel Palacios 14 Feb '07

14 Feb '07

On Feb 10, 2007, at 3:20 PM, Michael Stevens wrote: > My personal policy for my own stuff is to put the ticket on the > bugtracker and mail -dev to let people know it's there. This seems to > work. > > It'd be nice if there was something more organised. > > Michael When first moving to MacOSForge we setup a mailing list specifically for trac ticket activity, but such has never worked for reasons I don't exactly remember at the moment. I have a list of pending MacPorts infrastructure issues/requests I need to take up with MacOSForge admin personnel (which basically expands just to Kevin Van Vechten), and this one is certainly of top priority. As soon as I get a chance I'll make a post to this list detailing what I plan to bring up to Kevin's attention, so that others get a chance to point out anything I might have missed and to prioritize based on consensus. Regards,... -jmpp

1 0

NEW: arpack
by Mark Moll 14 Feb '07

14 Feb '07

Could someone commit this new port?: arpack, a package for solving large scale eigenvalue problems http://trac.macports.org/projects/macports/ticket/11279 -- Mark

2 3

NEW: xmlindent 0.2.17
by Cédric Luthi 14 Feb '07

14 Feb '07

Could someone please take care to commit this new port. http://trac.macports.org/projects/macports/ticket/11382 Thanks. Cédric Luthi

2 1

Re: [22007] portutil.tcl mods / more portfile.7 changes needed?
by Kevin Ballard 14 Feb '07

14 Feb '07

I will point out that if xinstall copies directories, that raises the question of how to apply mode and owner/groups to the installed directories. owner/group should be safe to recurse, but mode wouldn't be. On Feb 14, 2007, at 3:31 AM, Jordan K. Hubbard wrote: > On Feb 13, 2007, at 11:19 PM, Kevin Ballard wrote: > >> xinstall and file copy are different - xinstall is implemented in >> pextlib, and apparently doesn't recursively copy directories. It's >> a shame. >> >> So yeah, while xinstall won't handle directories, copy will do it >> just fine. The difference is that copy won't handle file >> permissions, but xinstall does. > > I was trying to copy the semantics of install(1) in order to obey > what seemed to be the principle of least astonishment. That said, > I'm having a hard time thinking of just how and where the world > would end if someone were to enhance it to actually copy > directories. I was probably being too conservative. -- Kevin Ballard http://kevin.sb.org eridius(a)macports.org http://www.tildesoft.com

2 1