True, but when adding an archive link with square brackets, these characters should be replaced by %5b and %5d because otherwise the link won't work.

Link Investigator is usually activated through the key F4. If a user attemps to refresh the page with F5 but instead accidentially presses F4, the tool is activated. Without a proper blocklist for Wikipedia this causes problems on watch lists, since the tool goes on to check (almost) every link.

@Cyberpower678 Thank you

Note how the page for https://nl.wikipedia.org/wiki/Speciaal:RecenteWijzigingenGelinkt/Sjabloon:Citeer_web contains no content. If you try with a target that has content, you should get a 200.

If you specify a target, i.e. https://en.wikipedia.org/wiki/Special:RecentChangesLinked?target=Main+Page, the HTTP response is a 200.

@Aklapper I tried with Postman on the same computer, and that returned HTTP 404.

I didn't see the parameter was double. Thanks for looking so closely. :)

How?

Screenshot, illustrating the problem

https://nl.wikipedia.org/w/index.php?title=Gebruiker%3AInternetArchiveBot%2FDead-links.js&type=revision&diff=49983274&oldid=49853706

Reverted unwanted changes: https://nl.wikipedia.org/w/index.php?title=Packet_sniffer&type=revision&diff=49893269&oldid=49892876

How can a wiki change the behaviour?

The arcive.org link is a good archive. Only the bot didn't find this link. Why not?

Please ignore 1, it works better now.

See https://phabricator.wikimedia.org/T172737

Declining number 1. There's a reason why the button disables when clicked on.
-> make it more clear then.

I would like to know WHY this URL was seen as dead. The false positive reporter doesn't tell me that.

We chatted about this, and we came up with an idea: if we can disable converting the encoding for existing archive.org URLs, I can live with this solution. I propose to call this "convert_archives_encoding=0" but feel free to change the parameter name.

You could afterprocess it, when it comes out of the external library. I will ask others for opinions on this matter.

When reading the URLs, you santize the URL to store it in the database. You write that sanitized URL back to Wikipedia, but with a few exceptions:

• use + instead of its encoded variant
• use = instead of its encoded variant (unless you really want to comply with RFC standards)

In T171859#3486925, @Cyberpower678 wrote:

The biggest reason is to keep the DB of IABot from cluttering with duplicates. The second reason is if the archive conforms to proper standards, it makes it easier for the Wayback Machine to load it, and should a change that breaks support for improper encoding, this URL is less likely to break from said hypothetical change.

In T171859#3486876, @Cyberpower678 wrote:

The first URL doesn't get encoded that way. = is only encoded in the path. This is done because = isn't actually legal in the URL and is only used to define values to parameters in the queries, based on RFC standards.

OK

https://google.be?query=test
to
https://google.be?query%3Dtest

I still have this. I used the Dutch Wikipedia to login. I tried Chrome as well, and as a related issue I get the same error for an URL that has already been whitelisted when I click the blue SUBMIT button.

We'll see whether it stays :)

In T170739#3441275, @Cyberpower678 wrote:

You are a sysop there, you have the means to do it.

In T170739#3441264, @Cyberpower678 wrote:

Wouldn't it be simpler if this was linked from the page history like on enwiki?

That would be nice as well.

My mistake, it seems to work now. I will turn on InternetArchiveBot again.

I have disabled InternetArchiveBot for now. This is too important. On the Dutch Wikipedia, there are a lot of archive.is links like this.

@Cyberpower678
Yes, thank you for that.

What @Effeietsanders meant to say:

@Cyberpower678
What @Effeietsanders meant by his comment:

In T136148#3422103, @Effeietsanders wrote:

@Smile4ever that would be a workaround, but rather undesirable. After all, the link still works fine - and changing it to yes might trigger it to be reported as dead in the database. It may very well be that we only want to use this specific archive version only on that article. For example: in case of a government page with the makeup of the cabinet. In some articles A we may want to refer to a specific version (the page as it was in 2007), in some other articles B to another version (the page as it was in 2010) and in some other articles C yet, we may want to refer to the most up to date version (the current version). If I report it as dead in A, I fear that it will eventually trickle also to B and C. It would probably be better to have some kind of override (archiveonly = yes, for example) that we would only use if we really want to point to a very specific version.

In T169652#3405134, @Cyberpower678 wrote:

I'm actually confused here. Your first sentence is asking me to use "nee", but your third sentence states using "nee" all the time is a problem. Can you please clarify your intentions?

In T136148#3404296, @Effeietsanders wrote:

IABot NEVER replaces archives with the original URL. It's not programmed to do that. I need an example of what you mean.

Examples:

• <s>https://nl.wikipedia.org/w/index.php?title=Antillenhuis&diff=prev&oldid=49391417</s>
• https://nl.wikipedia.org/w/index.php?title=Antipsychiatrie&diff=prev&oldid=49391443
• https://nl.wikipedia.org/w/index.php?title=Antari&diff=prev&oldid=49391285

https://nl.wikipedia.org/w/index.php?title=Adware&diff=next&oldid=49358122

Order is incorrect. Plural for "bronnen" is also incorrect. See https://nl.wikipedia.org/w/index.php?title=...Something_to_Be_%28single%29&type=revision&diff=49365878&oldid=37572390