RSS feed is not valid

[nilcolor+trac@…]

I have an Trac 0.11b1 installation and want to subscribe to RSS feed for tickets i need to monitor. But RSS feed is not visible within NetNewsWire Lite (v3.1b4). But its visible in Vienna (v.2.2.2.2212)… So i make a decicion that RSS feed have some "not-fatal" error within ('cause NNW is know as strict RSS reader) while Vienna allow and display this feed. I can't check feed with online validator - i have intraweb Trac so can't say exactly where is error is. Feed from this site is displayed well in NNW (i try this link ​http://trac.edgewall.org/report/20?format=rss&USER=anonymous)

[Emmanuel Blot]

Weird. I'm using NetNewsWire 3.0/1130 and I have no trouble getting feed from a Trac 0.11b1 server, neither from the changelog nor from the report view.

Report {20} is not part of the pristine Trac distribution. Which RSS URL is actually in trouble ?

[nilcolor+trac@…]

Firefox and NetNewsWire Lite screenshot with report {1} as RSS

[nilcolor+trac@…]

I try report {1} (All Tickets) as well as report {7} (My Tickets) - both didn't work within NNW and works with Vienna/Firefox And i try RSS feed from query like this ​http://portal/trac/tsm/query?status=accepted&status=assigned&status=new&reporter=nilcolor%40gmail.com&order=priority&col=id&col=summary&col=status&col=owner&col=type&col=priority&col=milestone That RSS is not working in NetNewsWire Lite 3.1b4 and its ask me to save .xml file if i paste that url in FireFox…

[Emmanuel Blot]

Replying to nilcolor+trac@gmail.com: http://portal/trac/tsm/query?status=accepted&status=assigned...

This is not a valid RSS feed URL: Trac RSS feed URL always contains a "format=rss" parameter. Please double-check this URL.

I copied 'n pasted your URL with a local installation, added the "format=rss" parameter, and obtained a feed that NetNewsWire has been able to read with no trouble. So either your URL is wrong or some of the cyrillic characters are causing some trouble.

Please retrieve the content of an invalid RSS feed with wget and attach it to this ticket. Use the --save-headers option switches so that the attached file contains all the HTTP headers.

[anonymous]

Sorry, i didn't have wget - and i download feer with next commands:

curl --include "http://itportal/trac/tsm/report/1?format=rss&USER=anonymous" --ou
tput report-1.rss

and

curl --include "http://itportal/trac/tsm/query?status=accepted&status=assigned&st
atus=new&format=rss&order=priority" --output query-1.rss

Both files attached…

[nilcolor+trac@…]

/query and /report/1 result feeds (archived)

[Emmanuel Blot]

The RSS you've attached is not valid. One issue is already known (multiple pubDate), the other issues may need some fixing. Stay tuned…

[Emmanuel Blot]

Patch proposal to fix up invalid RSS report feed.

[Emmanuel Blot]

Could you try the attached patch file (report-6556.diff​) and check if the report RSS feed works for you? Note that this patch does not address query feeds, only report feeds.

[osimons]

I haven't reveiwed or tested the patch, but just spotted that you used this to get the logo url:

chrome.logo.src_abs and chrome.logo.src_abs or abs_href(chrome.logo.src)

It won't work. If logo url is http://... it will print 'True' in the template - useful :-) If it is a regular project reference, it already gets calculated inside trac.web.chrome.Chrome.get_logo_data() so what you then abs_href() is already href()'ed - meaning the path to the project gets duplicated (like http://.../trac/project/trac/project/chrome/...).

I added a ticket for this (see #6562) - it would be OK if you could use the opportunity to see how that works with your rss modifications. A second opinion would be very useful. Until that gets committed, the best alternative is just sticking to the single line with <url>$chrome.logo.src</url>. I'll hopefully update that to something better soon.

[anonymous]

Replying to eblot:

Could you try the attached patch file (report-6556.diff​) and check if the report RSS feed works for you? Note that this patch does not address query feeds, only report feeds.

i apply .diff file and restart webserver (Apache). But NetNewsWire still didn't show anything. New feed file attached.

[nilcolor+trac@…]

RSS file after applying patch

[Emmanuel Blot]

Replying to anonymous:

i apply .diff file and restart webserver (Apache). But NetNewsWire still didn't show anything. New feed file attached.

Well, the W3C RSS feed ​validator reports: "This is a valid RSS feed." with your latest attachement, so at least the feed is valid. Can you check whether the error window in NNW reports any useful information?

[Emmanuel Blot]

Replying to osimons:

It won't work. If logo url is http://... it will print 'True' in the template - useful :-) If it is a regular project reference, it already gets calculated inside trac.web.chrome.Chrome.get_logo_data() so what you then abs_href() is already href()'ed - meaning the path to the project gets duplicated (like http://.../trac/project/trac/project/chrome/...).

I did not check how these routines behave in details, so I've probably missed an important point here. The trouble is that the URL produced with <url>$chrome.logo.src</url> is invalid as it contains only the server part, which is not valid for feeds (contrarily to HTML pages). I'll try your proposal (#6652) and report the result back. Thanks for the hint.

[anonymous]

Replying to eblot:

Well, the W3C RSS feed ​validator reports: "This is a valid RSS feed." with your latest attachement, so at least the feed is valid. Can you check whether the error window in NNW reports any useful information?

It works now… Really. Maybe it was some sort of cache problem but now NNM Lite shows me tickets. Thanks for fix. And thanks for Trac!

[Emmanuel Blot]

Replying to anonymous:

It works now… Really. Maybe it was some sort of cache problem but now NNM Lite shows me tickets. Thanks for fix. And thanks for Trac!

As osimons reported, the fix introduces a regression with the logo URL and needs to be fixed. At least the other issues seem to have been addressed. Can you verify the news date are valid - i.e. matches the one in your reports?

[nilcolor+trac@…]

Replying to eblot:

As osimons reported, the fix introduces a regression with the logo URL and needs to be fixed. At least the other issues seem to have been addressed. Can you verify the news date are valid - i.e. matches the one in your reports?

It seems that dates is OK. One example. I create one ticket and:

• Trac tell me that it was created 2007-12-28T16:34:22Z+0300
• Within RSS file its <pubDate>Fri, 28 Dec 2007 13:34:22 GMT</pubDate>
• And NNW shows me 28/12/07 16:34

So its seems that date is OK. I live in Russia, Moscow and my timezone GMT+3

[osimons]

Chatted with eblot, and we decided to bump the priority on this - valid RSS feeds should be 0.11.

[nilcolor+trac@…]

Nice to hear. Will wait for 0.11. And thanks for help!

[Christian Boos]

I didn't notice the patch at first. attachment:report-6556.diff​ looks good to me, could you please apply it manu if this actually fixes the issue?

[nilcolor+trac@…]

Hm.. Just upgraded to Leopard (not sure it means something but…) and install fresh copy of NNW (its free now BTW). And Active Ticket RSS stop working again.

• Online RSS verificator says its OK. (One warning though)
• NNW Error log shows nothing
• But feed is empty (i have active tickets - i'm sure) and market gray within NNW
• Safari/Vienna show feed OK. NNW - not.

[Thijs Triemstra <lists@…>]

Works fine with NNW 3.27 on Leopard 10.5.8 for ​http://trac.edgewall.org/report/20?format=rss&USER=anonymous here.

[Thijs Triemstra <lists@…>]

The RSS validator ​reports 2 warnings:

• line 17, column 29: Image title doesn't match channel title
• line 125, column 2: Missing atom:link with rel="self"

Perhaps this ticket can be about fixing those.

[Thijs Triemstra <lists@…>]

Here's a patch that should fix the first warning:

trac/ticket/templates/report_list.rss

@@ -8 +8 @@
     <description>List of available reports</description>
     <language>en-us</language>
     <image py:if="chrome.logo.src_abs">
-      <title>$project.name</title>
+      <title>$project.name: Available Reports</title>
       <url>$chrome.logo.src_abs</url>
       <link>${abs_href.report()}</link>
     </image>

[Christian Boos]

No, the channel title is different for each report (here you were just testing the list of all available reports, which is itself a report).

[Christian Boos]

Any chance for an updated patch? "Unschedul'ing" for now.