edesiocs / wub Goto Github PK

What steps will reproduce the problem?
1. Use static html file with cyrillic utf8

What is the expected output? What do you see instead?
I expect russain text. But I see some strange symbols. 

What version of the product are you using? On what operating system?
Last. I'M using Ubuntu 14.10 and ActiveTcl 8.6.3


Please provide any additional information below.
1. The html file contains all meta tags  charset=utf-8 etc.
2. I've tried other http server and I've got correct view.
3. The output from listener is:
<-- begin
1416817969299-66mS log @@{Listener sock1926520 on -blocking 1 -buffering full 
-buffersize 4096 -encoding utf-8 -eofchar {} -translation auto -sockname 
{0.0.0.0 0.0.0.0 8083 :: :: 8083}}
1416817969299-0mS site @@{Loading HTTPS listener Https}
<-- end

extra switch pattern with no body
    while executing
"switch -glob -- "[dict get $r -host],[dict get $r -path]/" { {
        Http Redir $r [dict get $r -path]/
    }
        default {
        NotFound $r
        }
    }"
    (procedure "trailing" line 2)
    invoked from within
"trailing $r"
    (procedure "::Dispatcher::do" line 33)
    invoked from within
"::Dispatcher do REQUEST $r1"

sorttables without footer are not closed at the bottom. This is because 
sorttable required footers to be used. This patch will use the headers as 
footers when no footers are specified in dict2html:

    # dict2table - convert dict into sortable HTML table
    proc dict2table {dict header {footer {}}} {
    if {![llength $footer]} {
        set footer $header
    }
    set row 0
    return [<table> class sortable [subst {
        [<thead> [<tr> [Foreach t $header {
        [<th> [string totitle $t]]
        }]]]
        [If {$footer ne {}} {
        [<tfoot> [<tr> [Foreach t $footer {[<th> [string totitle 
$t]]}]]]
        }]
        [<tbody> [Foreach {k v} $dict {
        [<tr> class [If {[incr row] % 2} even else odd] \
             [Foreach th $header {
             [If {[dict exists $v $th]} {
                 [<td> [dict get $v $th]]
             } else {
                 [<td> {}]
             }]
             }]]
        }]]
    }]]
    }

    # dict2table - convert dict into sortable HTML table
    # provisional new version
    proc dict2table {dict header {footer {}} {tag ""}} {
    if {![llength $footer]} {
        set footer $header
    }
    set row 0
    return [<table> class sortable [If {$tag ne ""} { class $tag
 }] [subst {
        [<thead> [<tr> [Foreach t $header {
        [<th> class $t [string totitle $t]]
        }]]]
        [If {$footer ne {}} {
        [<tfoot> [<tr> [Foreach t $footer {[<th> [string totitle 
$t]]}]]]
        }]
        [<tbody> [Foreach {k v} $dict {
        [<tr> class [If {[incr row] % 2} even else odd] \
             [Foreach th $header {
             [If {[dict exists $v $th]} {
                 [<td> class $th [dict get $v $th]]
             } else {
                 [<td> {}]
             }]
             }]]
        }]]
    }]]
    }

Hello Colin

In your newest Httpd.tcl with the Thread-Part you have copy the method rdump to 
dump but the call in Suspend and Resume is allways rdump.

I don't now if you must change the call rdump also to dump or the method from 
dump to rdump.

Best wishes
Marc

For my Wub-App I must have a replace of an element in the Query.tcl

What you Think about that, or have you a other possibility?

Marc


    #20100920/mz
    # replace - replace a simulated query element
    # query - query dict generated by parse
    # el - name of element
    # val - value of element
    # metadata - metadata
    proc replace {query el val {metadata {}}} {
    if {[dict exists $query $el]} {
        dict set query $el $val
    } else {
        dict lappend query $el $val $metadata
    }
    return $query
    }

Trunk version and last release don't work both. Previous versions are too slow 
- about few queries per second on desktop hardware.

CGI generated content appears to be cached too avidly.

What steps will reproduce the problem?
1. svn version 2083 checked out 11/21/2009
2. any Direct object example from examples package run ex9 for instance. 
3. cd into ex9 directory 
4. on command line start wub "/opt/usr8.6b.1/bin/tclsh8.6 ex.tcl
5. on browser goto url http://localhost:<port>/directoo/

What is the expected output? What do you see instead?
expect to see : (note html extracted)

Default function for MyOODomain

    * /default
    * /test

Instead I see errorinfo : 


Can't find method /default in any class of Direct::"


What version of the product are you using? On what operating system?
svn version 2083 checked out 11/21/2009
Linux Fedora 9 

Please provide any additional information below.
I fixed the problem by modifying Direct.tcl see output of svn diff below
Behavior at least from presentation on wub given at the 2009 European
Tcl/Tk conference is to map / to /default so wild card was wrong ( see
first diff); It could be that the example is wrong in this case and the
method /default should be changed to / ? Another problem was that the
search for methods to invoke in the object was seaching [self] and not
$object so all the methods of self in this case Direct:: were being
searched this lead to the error. 
Index: Direct.tcl
===================================================================
--- Direct.tcl  (revision 2083)
+++ Direct.tcl  (working copy)
@@ -145,7 +145,7 @@
     # locate a matching direct method in an object
     method do_obj {rsp} {
    Debug.direct {do direct $object $mount $ctype}
-   
+
    # search for a matching command prefix
    set fn [dict get $rsp -suffix]
    set cprefix [file split [armour $fn]]
@@ -273,13 +273,13 @@
     constructor {args} {
    set ctype "x-text/html-fragment"
    set mount "/"
-   set wildcard /
+   set wildcard /default

    foreach {n v} $args {
        set [string trimleft $n -] $v
        Debug.direct {variable: $n $v}
    }
-
+   
    # one or the other of namespace or object must exist
    if {![info exist namespace] && ![info exists object]} {
        # no namespace or object - must be a mixin
@@ -316,7 +316,7 @@
        set mixins [info class mixins $class]
        foreach m [lreverse [lsort -dictionary [info class methods $class
-private -all]]] {
        if {[string match /* $m]} {
-           foreach class [list [info object class [self]] {*}$mixins] {
+           foreach class [list [info object class $object ]  {*}$mixins] {
            if {![set unfound [catch {
                lindex [info class definition $class $m] 0
            } def eo]]} {

What steps will reproduce the problem?
1. Install Polipo 1.0.1 with the disableVia=false option
2. Configure Web browser to use Polipo as HTTP proxy
3. Enter http://wiki.tcl.tk/ in the address bar

What is the expected output? What do you see instead?
I should see the Tcl wiki, but instead I get "The proxy on (host:port)
encountered the following error while fetching http://wiki.tcl.tk/:
504 Proxy loop detected".

I emailed the author of Polipo about this issue, and he responded with more
information about it:

> Accessing any page in the Tcl Wiki (http://wiki.tcl.tk/) results in an
> erroneous 504 when using Polipo 0.9.9.  Upgrading to 1.0.1 fixes the
> problem.

No, it doesn't, it just works around it.  You can still reproduce the
issue with 1.0.1 if you set

 disableVia: false

in Polipo's config file.

The Wiki code is broken.  For some strange reason, it is echoing the
client's Via header in the reply.  This breaks Polipo.

                                       Juliusz 

What steps will reproduce the problem?
1. Put a file named "15%20Y.txt in a directory associated with a File domain 
(e.g. "/dir/").
2. View the directory listing (i.e. "/dir/"). 
3. Click on the "15%20Y.txt" link.


What is the expected output? What do you see instead?

  I expect the contents of the file. I see an error message saying the file whose name I just clicked on doesn't exist. The name has had the percent sign and following digits replaced with the character encoded by the value of those digits.

What version of the product are you using? On what operating system?

  Wub-4.0.0 on MS Windows XP Professional

Please provide any additional information below.

  The URL decoding evidently interprets escape sequences properly. It seems that the URL link in the directory display merely does not escape characters that it needs to. Percent signs in the link are the only non-cosmetic problem I've seen, but HTML entities in the file name (e.g. é.txt) are understandably rendered as the corresponding character by the browser. Perhaps they should be "sanitized" ("armoured"?) for display.

expose mkdir somewhere

What steps will reproduce the problem?
1. start wub as in the QuickStart guide
2. in browser surf to any directoo url
3.

What is the expected output? What do you see instead?
Not an errorInfo output. 
ErrorInfo complaining about failure of Utilities/OO.tcl methods failing. 

What version of the product are you using? On what operating system?
running on Linux Fedora 9 using Tcl 8.6b.1 

Please provide any additional information below.
Code in OO.tcl was failing in two places because oo instrospection commands
were illegal/ill-formed. svn diff below of edits I made to make
it work again. 
Index: OO.tcl
===================================================================
--- OO.tcl  (revision 2083)
+++ OO.tcl  (working copy)
@@ -5,7 +5,9 @@
 package provide OO 1.0

 proc ::oo::Helpers::classvar {name args} {
-    set ns [info object namespace [uplevel 1 {self class}]]
+    # added by cjolly because following line failed
+    set ns [uplevel 1 { self namespace}]
+    #set ns [info object namespace [uplevel 1 {self class}]]
     foreach v [list $name {*}$args] {
    uplevel 1 [list namespace upvar $ns $v $v]
     }
@@ -13,7 +15,8 @@

 proc ::oo::define::classmethod {name {args {}} {body {}}} {
     set class [lindex [info level -1] 1]
-    set classmy [info object namespace $class]::my
+    set classmy [uplevel 1 { self namespace}]::my
+    #set classmy [info object namespace $class]::my
     if {[llength [info level 0]] == 4} {
         uplevel 1 [list self method $name $args $body]
     }

What steps will reproduce the problem?
1. add Repo domain with upload 1
2. ensure you have no more than 51284 KiB of free RAM and no swap
3. upload any file of 5653856 bytes length

What is the expected output? What do you see instead?

Expected is that file is uploaded. Instead, Wub dies with this:
unable to alloc 11307732 bytes
Aborted


What version of the product are you using? On what operating system?
Revision: 2546

Please provide any additional information below.
evening_trk.gpx --- file to upload
core.18075 --- tclsh8.6 core dump
backtrace --- tclsh8.6 backtrace

Repo can tar on download, so it makes no sense to upload tar.gz files - so
unpack them on upload.

Repo domain doesn't clean up its temp tar files when asked to produce a tar for 
a whole directory.

Although they are all emitted into /tmp, they still leak, and over time 
accumulate to the detriment of the system.

Problem is I don't know the lifetime of the generated files, have no means of 
scheduling a [file delete] when they're no longer useful.

One possible solution would be to generate them into a subdir, then 
periodically clean up everything older than some magic age.

What steps will reproduce the problem?
create a File domain, referencing a docroot
start up using chrome, or IE with google toolbar (haven't tested other browsers)
access a page in the file domain that doesn't exist.

What is the expected output? What do you see instead?
you should get a 404 with a page that says
File 'somefile.html' doesn't exist

What version of the product are you using? On what operating system?
svn latest.. 

Please provide any additional information below.

http://www.google.com/support/forum/p/Toolbar/thread?tid=06ca1ad3bf06b621&hl=en
and
http://stackoverflow.com/questions/3970093/include-after-php-404-header-returnin
g-oops-this-link-appears-to-be-broken

mostly seems to bother chrome, but if a 404 has < 512 characters, it will 
display a weird error page.  
This can be fixed by padding the 404 with extra characters.

(the stack overflow page has some nice ascii art that someone uses).

Here is a diff of Utilities/Http.tcl that will pad the content with enough 
spaces so the not default 404 message will work with chrome.


$ svn diff Http.tcl
Index: Http.tcl
===================================================================
--- Http.tcl    (revision 3673)
+++ Http.tcl    (working copy)
@@ -764,6 +764,12 @@
        dict set rsp -code 404
        dict set rsp -rtype NotFound

+       set content_length [string length [dict get $rsp -content]]
+
+       if {$content_length < 512} {
+               dict append rsp -content "\n<!--[string repeat " " [expr {512 - 
$content_length} ]] \n-->"
+       }
+
        return $rsp
     }

What version of the product are you using? On what operating system?

svn latest.  
Windows

Please provide any additional information below.

Site.tcl
change
  return [uplevel source_org {*}$args]
to
  return [uplevel source_org $args]

uplevel concats its arguments together.  using {*} causes it go get expanded 
twice.
If your filename has a space, it will throw an error.

There is another incorrect variable name error in Listener.tcl. It shows up
in the socket error when we try to bind a socket that is already in use.
ie: fire up on port 80 when tclhttpd is already using that port.
The following solves it:

/opt/wub $ svn diff
Index: Wub/Listener.tcl
===================================================================
--- Wub/Listener.tcl    (revision 2045)
+++ Wub/Listener.tcl    (working copy)
@@ -121,7 +121,7 @@

            Debug.listener {server: $cmd}
            if {[catch $cmd listener eo]} {
-               error "[dict get $args -host]:[dict get $args -port]
$listen\ncmd=$cmd"
+               error "[dict get $args -host]:[dict get $args -port]
$listener\ncmd=$cmd"
            }
            Debug.log {Listener $listener on [fconfigure $listener]}
        } error eo]} {
/

What steps will reproduce the problem?
1. Source the attached wub_deadlock.tcl into an interpreter.  This is a minimal 
subset of Wub with Httpd::pre/post/do overridden to return hard-coded data.
2. Run 'openload -h User-Agent "MSIE 5.0" -t http://localhost/' to test a 
response to a single query
3. Run 'openload -h User-Agent "MSIE 5.0" http://localhost/' and wait for the 
server to stop responding (usually doesn't take long).


What is the expected output? What do you see instead?
Server shouldn't stop responding.  

What version of the product are you using? On what operating system?
I can reproduce with wub-svn-20090716 and Wub-4.0.0.tar.gz on Tcl 8.6b1.1 (cvs 
2010/03/25) and tcllib 1.11.1.
Getting more current builds on Windows is not straightforward, so I'm hoping to 
see if anyone else can reproduce this before updating everything to the 
bleeding edge.

Please provide any additional information below.
The server often seems to recover eventually if no further client connections 
are made, but sometimes it has not recovered after 15+ minutes.  Running [Httpd 
reaper] will sometimes cause immediate recovery, but not always.

<b>What steps will reproduce the problem?<b>
1. I think any time a page is loading and an error occurs in my code

<b>What is the expected output? What do you see instead?<b>
My error, but a wub-error occurs and I have no idea how to fix it:

wrong # args: should be "error message ?errorInfo? ?errorCode?"
    while executing
"error"

It doesn't even specify in which file this happens or code around it so that I 
can find it. I've done a search, but it started taking up too much time. 

There is also an error:

No such variable 'quote' in "unset quoted"


<b>What version of the product are you using? On what operating system?<b>

Wub-4.something on MS Windows XP Professional

<b>Please provide any additional information below.<b>

What steps will reproduce the problem?
I downloaded the exampels from wubwikit and tried to run example 1.


What is the expected output? What do you see instead?
expected - a web age.
result.  - error page.
Error Code 'NONE'

extra switch pattern with no body
    while executing
"switch -glob -- "[dict get $r -host],[dict get $r -path]/" { {
        Http Redir $r [dict get $r -path]/
    }
        default {
        NotFound $r
        }
    }"
    (procedure "trailing" line 2)
    invoked from within
"trailing $r"
    (procedure "do" line 32)
    invoked from within
"do REQUEST [pre $r]"



What version of the product are you using? On what operating system?
did a svn checkout.revisoni 3345.

Please provide any additional information below.
$ svn log Nub.tcl | less
------------------------------------------------------------------------
r3258 | mcccol | 2010-08-17 19:44:36 -0500 (Tue, 17 Aug 2010) | 2 lines

* added default facility to redirect URLs which would match if there were a 
trailing / to that URL



did a diff between that and previous version.

and found this.
+    proc code_trailing {processed} {
+       upvar 1 domains domains
+       set switch ""
+       foreach {u d} $processed {
+           set url [join [lassign $u host] /]
+           Debug.nub {code_trailing: $u ($d)}
+           dict with d {
+               lappend switch "$host,$url*"
+           }
+       }
+       set switch [join $switch " -\n"]
+       append switch { {
+           Http Redir $r [dict get $r -path]/
+       }}
+       append switch {
+           default {
+               NotFound $r
+           }
+       }
+
+       # set up a selector to specify trailing/
+       set selector {"[dict get $r -host],[dict get $r -path]/"}
+       set body "Debug.nub {trailing: \[dict get \$r -host],\[dict get \$r 
-path]/}"
+       append body \n
+       append body "set result \[switch -glob -- $selector [list $switch]\]" \n
+       append body "Debug.nub {trailing: \[dict get? \$result -code] \[dict 
get? \$result location]}" \n
+       append body "return \$result"
+
+       Debug.nub {code_trailing: ($body)}
+       eval "::proc ::Httpd::trailing {r} [list $body]"
+    }
+

That code generates a bad switch body.
Quick glance and I couldn't see exactly what this code was trying to do.  

What steps will reproduce the problem?
1. add Repo domain with upload 1
2. ensure you have no more than 51284 KiB of free RAM and no swap
3. upload any file of 5653856 bytes length

What is the expected output? What do you see instead?

Expected is that file is uploaded. Instead, Wub dies with this:
unable to alloc 11307732 bytes
Aborted


What version of the product are you using? On what operating system?
Revision: 2546

Please provide any additional information below.
evening_trk.gpx --- file to upload
core.18075 --- tclsh8.6 core dump
backtrace --- tclsh8.6 backtrace

It seems that a stream of [fileevent] events can starve [after] events,
which presents a problem since Httpd uses [after] events to communicate
between coroutines.

Unsure of the root cause of this starvation, but it may be that we need to
limit the number of requests on a pipeline, or to introduce a break on the
pipeline's readable [fileevent] stream by occasionally setting the handler
to {}, ie: turning off the fileevent stream, then turning it on sometime later.

Probably need to measure the number of calculated but undelivered responses
to see if this is necessary.  That may be an analog of lag.

Best solution would be to make event handling starvation-free.

What steps will reproduce the problem?

Create a cgi script that displays all environment variables and access the 
url. The entire environment is made available. This needs restricting with 
a whitelist of variables necessary. From tclhttpd we can see on unix 
systems we require PATH, LD_LIBRARY_PATH and TZ. On Windows we will need 
some additional items which can be researched later.

The attached patch adds a 'whitelist' arg for the CGI domain so the set of 
safe variable names can be extended in nub files. I also fix a couple of 
the server generated variables to conform to the documentation.

Additionally, CGI scripts expect to receive the client provided headers as 
HTTP_VAR_NAME type variables. The code appears to do this by added a list 
of fields expected to be passed to the cgi. However, we need to pass any 
client headers. Sometimes we have script that expect additional headers 
from clients (say X-Bender: Kiss my shiny metal butt.) and this should be 
passed. To that end I added -clientheaders to be a list of fields that the 
client provided and the env command uses that to generate the variables.

Finally - as we can;t unset ::env, we remove all array items not in our 
whitelist and add all the new ones. Works on my system, works with 
gitweb.cgi.

What is the expected output? What do you see instead?


What version of the product are you using? On what operating system?


Please provide any additional information below.

What steps will reproduce the problem?
  wget
http://shadoura.com/osm/osm_button_blackwhite_inverted_front_back.svg
-O/dev/null -S 2>&1 |grep type

What is the expected output?

  content-type: image/svg+xml

or

  content-type: text/xml

What do you see instead?

  content-type: text/xml text/xml

What version of the product are you using? On what operating system?

  Revision: 2293, Debian GNU/Linux


Please provide any additional information below.

  Adding the line

image/svg+xml                  svg

into Utilities/Mime/mime.types and restarting Wub doesn't change anything

What steps will reproduce the problem?
1. Define a File domain
2. Place a file named "index.html" in a subdirectory named "just testing" 
in the directory for that domain
3. Try to navigate to the file, either via href or entering its URL

What is the expected output? What do you see instead?
  I expect to see the file displayed by the browser.
  I see instead the error message "File 'domain/just%20testing/index.html' 
doesn't exist"

What version of the product are you using? On what operating system?
  Running Windows XP professional.
  The /wub/ home page reports Tcl 8.6b1.1 and "Last modified: Fri Feb 27 
13:41:25 EST 2009" at the bottom.

Please provide any additional information below.

Hello Colin

Your Patch for the https has not workin it was coming the follwing error 
message:
3104-0mS error @@{accept: -code 1 -level 0 -errorcode NONE -errorinfo {unknown 
method "connect": must be Resume, Suspend, addSock, create, delSock, destroy, 
new or s2h\n    while executing\n"{*}[dict get $opts -httpd] $sock $ipaddr 
$rport {*}$opts -server $server"} -errorline 25}

I have now changed the call "Httpd connect" to "Httpd new", like you can see in 
the following patch. Is this good like this?

Marc

@@ -136,7 +136,7 @@ class create ::Listener {
    set args [dict merge [subst {
        -host [info hostname]
        -port 8080
-       -httpd {Httpd connect}
+       -httpd {Httpd new}
    }] $args]
    dict set args -id [self]

I've tried WubWikit for offline use, and found one issue in Httpd.tcl and a 
minor one in WikitWub.tcl (there are two places that create 
::tdbc::sqlite3::connection with -readonly 1; changing it to 0 and it will work 
fine).

The issue with Httpd.tcl is that in the method Resume, it calls its coroutine 
with two arguments. According to Tcl/Tk documentation, a coroutine can be 
resumed with optional single value. Calling it with two arguments caused an 
error.

To be exact, here is the line 2342 in file Httpd.tcl (version Feb 25, 2013):
set code [catch {{*}[dict get $r -send] send $r} e eo]

To fix the problem, I changed it to:
set code [catch {{*}[dict get $r -send] [list send $r]} e eo]

Hope this helps.

Wub contains a Cache which is intended to hold generated content for as
long as it's fresh, where the generator can evict stale content from the Cache.

This works really well, as long as there's some locality of reference
between page fetches.

Unfortunately, that is precisely what spider bots *don't* do.  They cut a
swathe through the URL space, seeking in effect to sample the freshness of
all available pages.

So bots evict Cache contents, willy nilly.

It is necessary to somehow distinguish bot access from human access, and to
prevent bots evicting perfectly usable content.  But how to do this? 
Enhance the Spider package to recognise *all* spiders, and make the
distinction between bad and useful spiders?  That's a lot of data, all
driven from User-Agent field.  How else?  Access pattern recognition?  Hard.

This problem is most evident in a Wiki, where Recent Changes tends to drive
traffic to the Cache.

The wiki seems to be treating /_/summary?N=1 as equivalent to
/_/summary?N=2 under caching.  This is clearly not what's expected.

What steps will reproduce the problem?

In Utilities/SMTP.tcl, in [sendmessage] method of the SMTP class. At line 472 
the code checks for missing message-id header, and if it's missing it generates 
and inserts it. At line 477, the code checks for missing content-type header, 
and if it's missing the code inserts a message-id header with value of 
"text/plain".

To connect to oure db_module we connect with a socket with async
datatransfer so we need vwait to serialize the query but in the Site
package you rename the vwait.

Why jou do that?
How can I make a client socket connection with vwait from the wub server to
oure db_module on a other server?

Thanks Marc

What steps will reproduce the problem?
1. check out latest version.
2. copy example.config to site.config
3. uncomment the "#nub" section from site.config
4. start the server.
5. browse to localhost:8080
6. Click on "server configuration".
7.  Click on "Add" in the nubs section.

What is the expected output? What do you see instead?

Expected to be prompted for userid/password in order to edit/add a nub.
instead I Get a static page that talks about the Nub domain.

The home page links to 
    http://localhost:8080/nub/nubs
A link from the static nub domain page links to
    http://localhost:8080/nub/nubs/

With a trailing "/", you do get prompted to login.



What version of the product are you using? 
version 3448

Please provide any additional information below.
I see a bunch of code dealing with URLs with trailing slashes.  I'm not sure if 
the main page is intended to link to nub/nubs/, or if the code is just supposed 
to work either way.

On a side note, 
Setting a password in the site.config in the nub section does not set the 
"password" var in Nub.tcl.

The latest commit for doing regsub substituted rewrites is broken in two 
places.
One is a missing ] from gen_definitions in the regsub segment after the 
::apply.
The other is 'no such variable' at line 886 in Nub.tcl where url has never 
been defined.

Cannot start Application.tcl without reverting to an earlier commit.

*What steps will reproduce the problem?*
1. mkdir "Dir with some spaces in the name"
2. wget http://wub-server/Dir%20with%20some%20spaces%20in%20the%20name

*What is the expected output?*
The directory listing.

*What do you see instead?*
404.

*What version of the product are you using? On what operating system?*
Latest SVN on GNU/Linux

*Please provide any additional information below.*
The patch is attached

What version of the product are you using? On what operating system?
svn latest - windows

Please provide any additional information below.

the errorInfo variable keeps getting overwritten with
wrong # args: should be "error message ?errorInfo? ?errorCode?"

I tracked this down to the watchdog timer.

     } elseif {$rc == [catch error ""]} {

I think you meant     
     } elseif {$rc == [catch {error ""}]} {


but I don't like that solution either.  It just populates errorInfo with 
something else.

I'd suggest either

     } elseif {$rc == 1} {

or if you really want to let tcl supply the code

     ...
     set temp_errorInfo $::errorInfo
     ...
     } elseif {$rc == [catch {error ""}]} {

     ...
     set errorInfo $temp_errorInfo


but I think the first approach is a lot easier.

Arjen reports that clicking on the 'View' link for an uploaded zip file gives 
garbage (actually, it sends the entity encoded as text/plain, which is much the 
same thing)

It would be better to not try to construe things as text/plain when they 
obviously can't be construed that way.

It would also probably be better to unpack uploaded zip, tar etc files and 
arrange their elements in the upload directory, which would obviate Arjen's 
problem.

What steps will reproduce the problem?
1. start Application.tcl
2. try to find stx.stx
3.

What is the expected output? What do you see instead?
* it is not under wub/docs/Utilities/stx.stx
* it is under wub/doc/Utilities/stx.stx

I can remember how I navigated to the correct page ... can repro now.


What version of the product are you using? On what operating system?
Revision: 2142


Please provide any additional information below.

What steps will reproduce the problem?

1. Open Wub URL: Domains/stx.stx
2. Look at 'Form Example 2'
3. Ditto for 'Form Example 3'

What is the expected output? What do you see instead?

Unbroken form. Broken one.

What version of the product are you using? On what operating system?

Wub SVN revno 1159, wubwikit-20081212.kit 


Please provide any additional information below.

It seems for me that above description is quite enough. A 2 cents medicine
is provided.