Upgrading

Migrating from 9.x to 10.0

Python 3.6 or newer is required.

The Python Beautiful Soup package is now required. A C compiler is not needed for building.

The loginuserfield and loginpasswordfield entries in the authentication section of linkcheckerrc are now matched to a login form case-sensitively.

The “blacklist” logger has been renamed to “failures”. “blacklist” is recognised as an alias for “failures” but will be removed in future.

Migrating from 8.x to 9.0

The Python requests package is now required.

Several checks have been moved to plugins (see below). Plugins have to be enabled in the configuration file.

The following commandline and configuration options have been deprecated and do not have any effect: –anchors, anchors: moved to plugin AnchorCheck –check-css, checkcss: moved to plugin CssSyntaxCheck –check-html, checkhtml: moved to plugin HtmlSyntaxCheck –complete: feature removed –cookies, sendcookies, storecookies: cookies are sent/stored per default –pause, wait: replaced with numrequestspersecond –scan-virus, scanvirus: moved to plugin VirusCheck –warning-regex: moved to plugin RegexCheck –warning-size-bytes, warnsizebytes: feature removed warnsslcertdaysvalid: moved to plugin SslCertificationCheck

The “html” logger generates HTML5 documents now.

The following warnings have been removed: - http-auth-unauthorized: removed - http-auth-unknonwn: removed - http-decompress-error: removed - http-robots-denied: downgraded to info - http-moved-permanent: downgraded to info - http-unsupported-encoding: removed - https-certificate-error: is an error now - mail-unverified-address: removed - mail-no-connection: removed - syntax-css: moved to plugin - syntax-html: moved to plugin - url-anchor-not-found: moved to plugin - url-content-size-unequal: removed - url-warnregex-found: moved to plugin

Migrating from 8.4 to 8.5

Custom output loggers have been changed. See doc/web/content/faq.md for an example with custom loggers.

Migrating from 8.0 to 8.1

All loggers have an additional output field “modified”. If these loggers are not configured with specific output parts, the output format will change. For example existing SQL tables can be altered with: alter table linkcheck add (modified varchar(256));

The default User-Agent string used by LinkChecker is now Mozilla/5.0 (compatible; LinkChecker/8.1; +http://linkchecker.sourceforge.net/)

Migrating from 7.9 to 8.0

Python 2.7.2 or newer is required (Python 3.x is not supported though).

Migrating from 7.6 to 7.7

The deprecated options –check-html-w3 and –check-css-w3 have been removed from the commandline client.

Migrating from 7.3 to 7.4

Python 2.7 or newer is required (Python 3.x is not supported though).

The deprecated options –interactive, –priority and –allow-root have been removed from the commandline client.

Migrating from 7.0 to 7.1

The FastCGI module lc.fcgi has been removed. The lc.cgi module can be used instead.

Migrating from 6.x to 7.0

The system configuration file support has been removed. There is now only one user-configurable configuration file. On Unix systems it is at $HOME/.linkchecker/linkcheckerrc and on Windows systems at %HOMEPATH%.linkcheckerlinkcheckerrc

Migrating from 6.6 to 6.7

The machine readable output formats xml, csv and sql now report the recursion level of each URL.

Migrating from 6.4 to 6.5

Only applies if a custom output loggers has been programmed: The URL data model for output loggers has changed. The list of warning messages is now a list of tuples (tagname, warningmessage).

Migrating from 5.5 to 6.0

Python 2.6 or newer is required (Python 3.x is not supported though).

The deprecated –no-proxy-for and –no-anchor-caching options have been removed.

The configuration file now requires multiline syntax for the options “nofollow”, “ignore” and “entry”.

Migrating from 5.2 to 5.3

The –password option now reads a password from stdin instead taking it from the commandline. This prevents reading the password from the commandline string with tools like ps.

Migrating from 5.1 to 5.2

The –no-proxy-for option has been deprecated and will be removed in a future release. Also, the “noproxyfor” entries in configuration files will not be evaluated anymore. You should use the $no_proxy environment variable instead, which specifies a comma-separated list of domains that are not contacted over proxies.

Migrating from 5.0 to 5.1

The –no-anchor-caching option has been deprecated and will be removed in a future release. This option should not be used anymore.

The exit code of the linkchecker script will be zero now if all warnings have been ignored.

Migrating from 4.x to 5.0

Python >= 2.5 is now required.

The CGI script access control has been removed. Please use the access control of your webserver to restrict access to the CGI script. An example configuration file for the Apache weberver has been included in the distribution.

Migrating from 4.4 to 4.5

Configuration file entries that were enumerated (ignoreX, nofollowX, noproxyX and entryX) now have a new multiline syntax. For example the old configuration:

ignore1=^mailto:
ignore2=^nntp:

should be written as:

ignore=
  ^mailto:
  # this is a comment
  ^nntp:

Note the leading spac(es) at the beginning of each line after ignore=. If an indented line starts with a comment sign, it will be ignored.

Migrating from 4.2 to 4.3

The deprecated --disable-psyco option has been removed.

Migrating from 4.1 to 4.2

If run under Unix system as the ‘root’ user, LinkChecker drops privileges and runs under the ‘nobody’ user account. The new option –allow-root prevents this, ie. enables the old pre-4.2 behaviour.

Migrating from 3.x to 4.0

The Python API of the linkcheck module has changed. If you were using the library directly instead of the commandline or CGI interfaces, you will have to adjust your code.

The default intern pattern matches now both http and https. When checking a site http://example.com/, all URLs of the form https://example.com/ will now also be checked recursively.

LinkChecker now honors a “Crawl-delay” entry in robots.txt files. The delay makes LinkChecker pause between requests to the corresponding server, so your checking time might increase if the server enforces such a delay through its robots.txt file.

URLs with invalid syntax are now cached, and they get matched now by the –ingore-url option.

Migrating from 3.0 to 3.1

The xml output logger has been renamed to gxml.

Migrating from 2.x to 3.0

The –warnings option is deprecated since warnings are now printed per default. A new –no-warnings has been added if one wants to have the old behaviour. Additionally, some old warnings about have been removed.

The previously deprecated –status option has been removed.

The options –intern, –extern and –extern-strict have been replaced by –ignore-url and –no-follow-url.

The configuration file format has changed. See the distributed linkcheckerrc default config for the new syntax.

Migrating from 2.2 to 2.3

The per-user config file is now ~/.linkchecker/linkcheckerrc (previous location was ~/.linkcheckerrc ).

The default blacklist output file is now ~/.linkchecker/blacklist (previous location was ~/.blacklist).

Python >= 2.4 is now required.

Migrating from 1.x to 2.0

The –output and –file-output parameters can specify the encoding now. You should check your scripts if they support the new option syntax.

Some added checks might trigger new warnings, so automated scripts or alarms can have more output than with 1.x releases.

All output (file and console) is now encoded according to a given character set encoding which defaults to ISO-8859-15. If you relied that output was in a specific encoding, you might want to use the output encoding option.

Migrating from 1.12.x to 1.13.0

Since lots of filenames have changed you should check that any manually installed versions prior to 1.13.0 are removed. Otherwise you will have startup problems.

The default output logger text has now colored output if the output terminal supports it. The old colored output logger has been removed.

The -F option no longer suppresses normal output. The old behaviour can be restored by giving the option -onone.

The –status option is now the default and has been deprecated. The old behaviour can be restored by giving the option --no-status.

The default recursion depth is now infinite. The old behaviour can be restored by giving the option --recursion-level=1.

The option --strict has been renamed to --extern-strict-all.

The commandline program linkchecker returns now non-zero exit value when errors were encountered. Previous versions always return a zero exit value. For scripts to ignore exit values and therefore restore the old behaviour you can append a || true at the end of the command.