Content-Disposition header.
- When downloading/referring to a particular patch embed the patch
number with the PR instead of using two different CGI variables.
This means that there aren't "&" in the patch URL's anymore which
makes things simpler for sh(1) like shells, by not requiring quoting
of URLs when using a command line tool, such as fetch(1), to
download patches.
- In HTML-ifying of URL's, handle when a URL contains a PR like string
which can e.g. be seen in PR www/113071.
- Add a comment about how to test script via the command line.
Submitted by: edwin, shaun
PR: www/113071
where no page was returned to the submitter resulting in many
duplicate PR's.
The pipe to sendmail was just being closed without the nice ".\n" to
tell sendmail this is the end of the mail. Now we always send "\n.\n"
to be sure the is a standalone "." on a line.
Why this wasn't a problem before, but is now, I'm not really sure, but
it's probably a perl 5.0 -> 5.8 change which triggered this.
If there are still problems after this commit, please report including
PR number(s).
Prodded by: bugmeister(s)
Co-prodding and testing help: edwin
Not fixed for far too long by: simon
own CVSweb so there is not really any big reason for us to have CVSweb
for them as well. [1]
Remove the "learn" CVS repository. It hasn't changed in 9 years... [1]
Remove the German and San Marino CVSweb mirrors from the list since
neither one has a hostname which resolves.
Suggested by: wosch [1]
the relevant parts onto CVSweb 3.0.6 stock files, so it is not entirely
impossible that some local changes were lost, though I don't think/hope
so.
The use of cgi-style.pl was merged by danger@. He also, together with
Dan Langille, got CVSweb 3 tested against FreshPorts prior to commit, so
this update should not blow up FreshPorts. Thanks to both for the help.
Note that this now means that the FreeBSD CGI scripts more or less
require perl 5.8 (or at least something newer than 5.0), to run without
local modifications - but people really shouldn't be running perl 5.0
these days anyway... (no comments on the fact that www.FreeBSD.org did
this until ~2 hours ago... ).
which has more strict taint checks.
This is a temporary stopgap while upgrading to perl 5.8 and migrating
to cvsweb 3 to get around the catch-22 that cvsweb 3 requires perl
5.8, and perl can't be upgraded to 5.8 if it breaks cvsweb.
- /faq.html -> /help.html
- merge /copyright.html into help.html
- section title without link and logo
- no logo in footer
- minor bugfixes in layout & design
- remove 'Quick Reference Categories'
This work included:
- move a few functions out of query-pr-summary.cgi into a new file -
query-pr-lib.pl, which now provides a procedure to display forms
- include query-pr-lib.pl in both query-pr.cgi and query-pr-summary.cgi
scripts
PR: www/35647
Approved by: trhodes (mentor), keramida (mentor)
Silence from: www@
couldn't specify full PR ID, e.g., bin/12345, in the
"PR number" field of the web form because the slash
was encoded as %2F by many browsers. Of course,
decoding the query is a must for a cgi in general.
- Also allow queries like query-pr.cgi?q= and query-pr.cgi?cat/num
- Fix comment
- Display query form when no results found
- Allow some other characters to be part of the URL [1]
PR: www/103628 [1]
Reviewed by: simon, remko
Approved by: keramida (mentor)
of generating full list of all open PRs in GNATS. I hope this change
will at least a bit decrease the load of GNATS.
Reviewed by: www@
Approved by: keramida (mentor)
- A multi-line mail header wrapping on a space wasn't handled correctly.
- My attempt at finding the end of patches with no delimiter inside
replies wasn't working in some cases; I've disabled it for now.
This was written from scratch, and offers numerous enhancements over the
old script; notably:
- Patches are parsed, coloured, uudecoded, and downloadable.
- Cleaner, clearer layout.
- Easier-to-read Audit-Trail.
Reviewed by: freebsd-www (well, I tried...)
Approved by: simon
can find the source via cvsweb in any case.
- Use $hsty_base more places for better compatibility with
cgi.FreeBSD.org (IE. when the static FreeBSD.org web pages are
somewhere other that the site running the CGI's).
fixes use of the script with Perl 5.8.
This should match all characters of current ports. If somebody finds
some new characters to use in a port name the matching regex will need
to be updated.
absolute base path, do the check case insensitive to handle if
HTTP_HOST e.g. comes in as FreeBSD.org.
Break a rather long line (being changed anyway) to improve
readability.
http://www.w3.org/1999/xhtml, which is the namespace used by the XHTML
DTD's.
This fixes layout/rendering of output from the CGI scripts when using
Opera.
confirm-code.cgi contains a preconfigured list of databases and their
parameters. When a request comes in, the database in the request's 'db'
parameter is checked for validity, and a code is generated, stored in
the appropriate database and returned.
Use this new script in send-pr.sgml and remove sendpr-code.cgi which is
now superceded.
stopped checking in INDEX into CVS long time ago. Last useful releases were
4.10 and 5.3.
- Remove code to display new/updated ports, it relies on periodical INDEX
updates in CVS (which we don't do anymore) and was disabled long time ago
anyway.
to permit entering the code in lowercase (since the generator doesn't
generate lowercase letters).
Use a relative url to querypr-code.cgi so that mirrors (and test boxes)
can use the query-pr cgis unmodified.
The test for expired codes was inverted, so emptied the database every
time the test ran. Due to the time that the sweep was made, this bug
only hit the user if someone else had submitted a PR in the period
between their obtaining the code and submitting the PR.
This is likely the cause of the unexplained failures that some users
were seeing, which we all blamed on misbehaving caches.
that be an error or success), as it's not currently easy to find your
way back there. Suggested in:
PR: www/88473
Submitted by: Roman Neuhauser <neuhauser at sigpipe dot cz>
default. This has been a little wrong ever since these scripts moved to
docs.freebsd.org, but became very wrong when the style sheets were added
since docs.freebsd.org doesn't serve them.
1. If we're not running under cgi (for the mailindex run out of cron)
2. If we're running on docs.freebsd.org (because it doesn't have the
style files, and the "home" link would point to docs home, not
the home home)
3. If we're running on people.freebsd.org, to facilitate my testing
Otherwise, keep it relative for the benefit of mirrors.
docs problem pointed out by: Mark Trettin (mkt AT trettin DOT org)
It looks like Firefox applies the relative size twice, making the fonts
too small. Instead, now they'll be too big - just a bit bigger than
the "Large" web site font.
This only applies to HTML4 pages, which as I mentioned before are
cvsweb and mailindex, and I don't think that mailindex uses tables.
This means that this problem should be fixed when someone (no promises)
imports cvsweb 3 which uses XHTML.
This commit brought to you by CSS, Web Quirks, IE and our favorite IRC channel.
Add an optional 2nd argument to html_header(); if it's missing or
supplied and non-zero, html_header returns XHTML doctype; if it's
supplied and zero, html_header returns HTML 4.01 doctype.
cvsweb and mailindex are the only HTML4 holdouts. cvsweb should
be updated to 3.x which is XHTML, but that's a seperate step.
I left $hsty_home alone, because I couldn't find anything that used
it or any obvious image to use for it in the new layout.
and all the translations caught up. Except for es/support.sgml and
es/publish.sgml, but they are old. This also fixes a cleanup for
ceri which he started long time ago.
making it hard to copy/paste the originator field for PR issues.
Also make query-pr.cgi xhtml compliant.
Submitted by: Daniel Gerzo <danger at rulez dot sk>
pass the xhtml compliant check, making us conforming to standards.
This also finishes the update to query-pr-summary.cgi which is now
XHTML 1.0 compliant.
Submitted by: Daniel Gerzo <danger at rulez dot sk>
easier to read when you are using a color capable browser.
This will not impact a text-only browser.
Changes include:
o Code cleanup
o Add CSS style's.
o Use the new CSS style's.
o Move double quotes to quotes where needed.
o Made the HTML (except for the header) html-compliant.
The usage of CSS makes it easy to merge later when/if needed.
Discussed on: www@
No objections from: www@
Submitted by: Daniel Gerzo <danger at rulez dot sk>
o Clean some tagging.
PR: www/82379
Noticed by: Ed Schouten <ed@fxq.nl>
Submitted by: Daniel Gerzo <danger@rulez.sk>
Reviewed by: simon
Approved by: trhodes (mentor)
of problem report viewed through the web interface of query-pr to
bug-followup@freebsd.org to reduce the diffs from the problem-reports
articles and the potential for confusing our users.
Submitted by: Julien Gabel <jpeg@thilelli.net>
Forced commit also to reg.cgi to note that it can be cvs rm'd once all
of {es,fr,ja,ru}/register.sgml have caught up.
Approved by: silence on doc-committers@
some bugs of the text-mode listings of the Gnats database:
- Shave 2 characters out of the submission date, by removing the
enclosing [] pair.
- Increase the width of the tracker field. It should now be wide
enough until PR standards/999999 (which is, hopefully, a very
long time).
- Decrease the width of Resp. to 10 characters, as the longer name
is now a bit shorter (after the removal of 'freebsd-' a while
back).
- Shorten the 'Description' a bit, to keep the entire listing at
less than 79 columns (which IIRC was broken by me in revision
1.45, during the split of HTML/text output in separate functions).
- Handle that 6.0 also uses .tbz packages.
- Simply refer to the -STABLE/-CURRENT branches by their major number
so we don't have to update that part of the script for each release.
- 5-STABLE currently points to packages-5-current since there is no
packages-5-stable yet.
- Change the default version to 5-STABLE/i386.
the last suggestion of Jie Cao from www/69780:
``3. If we strip the spaces at the beginning of the lines, we
can reduce the size with a further 90000 bytes.''
PR: www/69780
Submitted by: Jie Cao (gaoj (at) cpsc (dot) ucalgary (dot) ca)
database queries once again, because I've received complaints that the
HTML output mode of the PR query scripts outputs huge, bloated tables
after the style changes I made to beautify HTML output.
This moves all the visual tweaks of the output in a CSS stylesheet and
add a $t_style option to the common cgi-style.pl script. If not set,
the style defaults to empty, but users of cgi-style.pl can set this to
a local CSS style like this:
$t_style="<style type=\"text/css\">...</style>";
or use it to refer from the header of the HTML output to an external
CSS stylesheet.
The HTML output size is substantially reduced now (at least 50% of the
size was caused by many repetitions of style="" attributes), but it's
still larger than the preformatted <pre> output because of all the
<tr> and <td> tags.
<TITLE>FreeBSD Hypertext Man Pages: tunefs(8)</TITLE>
they should be in the form
<TITLE>tunefs(8)</TITLE>
When a user has half a dozen tabs open in their browser, the tabs all
read "FreeBSD Hy...". If they read "tunefs(8)" instead, the user
could navigate much more quickly.
submitted by: Carleton Vaughn <keebler@mindspring.com>
characters in the confirmation image.
en/gifs/1.pnm and en/gifs/I.pnm will be safe to remove 45 minutes after
the next web build and not before, so I'll do that tomorrow.
PR: www/69676
- Use a <table> to align data for HTML output similar to the one used by
CVSweb to list directory contents. This helps the eye to pick the
right line to click on when picking PRs to read from a long list.
Approved by: ceri, linimon, simon, silence on -www
CVS, and use this by default. The CVS version is still used if the
local updated version isn't readable. On www.FreeBSD.org this means
that the updated INDEX(-5) found on
http://www.FreeBSD.porg/ports/INDEX(-5) are used.
In essence this mean that this CGI script is actually useful
again since it's not returning out-of-date results.
- Disable the "new ports" HTML since it just doesn't work without
frequent INDEX versions in CVS. The support code for this isn't
removed at this to avoid breaking stuff, but should be removed
later.
Testing opertunity provided by: kuriama
related to querying GNATS together, in order to aid with the upgrade to
GNATS 4.
Switch over dosendpr.cgi and query-pr.cgi to use it.
query-pr-summary.cgi to follow.
