------0LJ696WDRL04RTIYJRUHHL1J6RNDV4
Content-Type: text/plain;
charset=utf-8
Content-Transfer-Encoding: quoted-printable
Hi,
I think I can work on the uscan option. Just tell me where to find the example.
Nb: I'll be in Brest on Tuesday
Best regards,
Xavier
Le 10 juillet 2025 23:21:43 GMT+02:00, gregor herrmann <
[email protected]> a écrit :
Since a couple of days, MetaCPAN changed their ways of fighting AI scrapers, according to [0] with some service called "signalsciences". Which, according to [1] works fine, except that it means that all requests from the commandline / programmatically /
with a non-javascript-capable browser don't work anymore. [2] This includes uscan(1). [3]
So this means that we won't do any updates of Debian packages representing CPAN distributions anymore.
As for potential fixes/workarounds:
- Setting a different User-Agent doesn't change anything.
- Bas found a way to change debian/watch to use the MetaCPAN API, cf. [4]. In my POV, that's both technically brilliant and aesthetically difficult :)
In general I think that updating ~5000 d/watch files (and some tools) is not an attractive option …
- Hacking uscan to mangle metacpan.org URLs (or to work around the javascript requirement from signalsciences in general) might be doable but that would have to go into stable and whatever as well.
- Using a redirector for uscan might work but probably also needs updating all d/watch files.
I guess talking to the MetaCPAN folks would be a reasonable next step. Is there anybody, maybe wih connections to them, willing to do this? I admit that I don't have the energy to put another task on my TODO list.
Or maybe someone else has other ideas …
Cheers,
gregor
[0] https://github.com/metacpan/metacpan-web/commit/212a28303f761c181a4fcd94d8b17e70650ac1c2 [1] https://blogs.perl.org/users/dean/2025/07/metacpan-running-really-well-thanks.html
[2] try
wget -O- https://metacpan.org/release/ack
curl https://metacpan.org/release/ack
GET https://metacpan.org/release/ack
[3] uscan --report -vvv
in a soure package directory
[4] https://salsa.debian.org/perl-team/modules/packages/libpdl-vectorvalued-perl/-/commit/fc92473
--
.''`. https://info.comodo.priv.at -- Debian Developer https://www.debian.org : :' : OpenPGP fingerprint D1E1 316E 93A7 60A8 104D 85FA BB3A 6801 8649 AA06 `. `' Member VIBE!AT & SPI Inc. -- Supporter Free Software Foundation Europe
`-
------0LJ696WDRL04RTIYJRUHHL1J6RNDV4
Content-Type: text/html;
charset=utf-8
Content-Transfer-Encoding: quoted-printable
<html><head></head><body><div dir="auto">Hi,<br><br>I think I can work on the uscan option. Just tell me where to find the example.<br><br>Nb: I'll be in Brest on Tuesday <br><br>Best regards, <br>Xavier<br></div><br><br><div class="gmail_quote"><div dir=
"auto">Le 10 juillet 2025 23:21:43 GMT+02:00, gregor herrmann <
[email protected]> a écrit :</div><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<pre class="k9mail"><div dir="auto">Since a couple of days, MetaCPAN changed their ways of fighting AI scrapers, according to [0] with some service called "signalsciences". Which, according to [1] works fine, except that it means that all requests from
the commandline / programmatically / with a non-javascript-capable browser don't work anymore. [2] This includes uscan(1). [3]<br><br>So this means that we won't do any updates of Debian packages representing CPAN distributions anymore.<br><br>As for
potential fixes/workarounds:<br>- Setting a different User-Agent doesn't change anything.<br>- Bas found a way to change debian/watch to use the MetaCPAN API, cf. [4]. In my POV, that's both technically brilliant and aesthetically difficult :)<br>
In general I think that updating ~5000 d/watch files (and some tools) is not an attractive option …<br>- Hacking uscan to mangle metacpan.org URLs (or to work around the javascript requirement from signalsciences in general) might be doable but
that would have to go into stable and whatever as well.<br>- Using a redirector for uscan might work but probably also needs updating all d/watch files.<br><br>I guess talking to the MetaCPAN folks would be a reasonable next step. Is there anybody,
maybe wih connections to them, willing to do this? I admit that I don't have the energy to put another task on my TODO list.<br><br>Or maybe someone else has other ideas …<br><br>Cheers,<br>gregor<br><br><br>[0] <a href="
https://github.com/metacpan/
metacpan-web/commit/212a28303f761c181a4fcd94d8b17e70650ac1c2">
https://github.com/metacpan/metacpan-web/commit/212a28303f761c181a4fcd94d8b17e70650ac1c2</a> [1] <a href="
https://blogs.perl.org/users/dean/2025/07/metacpan-running-really-well-thanks.html">
https://blogs.perl.org/users/dean/2025/07/metacpan-running-really-well-thanks.html</a><br>[2] try<br> wget -O- <a href="
https://metacpan.org/release/ack">https://metacpan.org/release/ack</a><br> curl <a href="
https://metacpan.org/release/ack">https:
//metacpan.org/release/ack</a><br> GET <a href="
https://metacpan.org/release/ack">https://metacpan.org/release/ack</a><br>[3] uscan --report -vvv<br> in a soure package directory<br>[4] <a href="
https://salsa.debian.org/perl-team/modules/packages/
libpdl-vectorvalued-perl/-/commit/fc92473">
https://salsa.debian.org/perl-team/modules/packages/libpdl-vectorvalued-perl/-/commit/fc92473</a><br></div></pre></blockquote></div></body></html>
------0LJ696WDRL04RTIYJRUHHL1J6RNDV4--
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)