Broken links in index pages #12

Activity

1. timsong-cpp commented on Apr 19, 2026

   

   timsong-cpp

   on Apr 19, 2026

   Owner

   More actions

   The |see thing still exists upstream, I've reported Eelis/cxxdraft-htmlgen#123

   I'm not going to try to figure out how to patch historical versions of the generator tool, but I'll consider PRs that fixes the HTML.
2. jhcarl0814 commented on Apr 19, 2026

   

   jhcarl0814

   on Apr 19, 2026

   ContributorAuthor

   More actions

   @timsong-cpp Got it👌 Thanks!
3. jhcarl0814 commented on Apr 27, 2026

   

   jhcarl0814

   on Apr 27, 2026

   ContributorAuthor

   More actions

   The ids of 2 hash_functions are conflicting with the ids of 2 hash functions in n3337/generalindex, n4140/generalindex and n4618/generalindex. The 2 hash_functions should be moved to n3337/libraryindex, n4140/libraryindex and n4618/libraryindex, but unlike newer versions of the libraryindex pages, the libraryindex pages don't have a unordered associative containers section for the 2nd hash_function to move to.
4. timsong-cpp commented on Apr 29, 2026

   

   timsong-cpp

   on Apr 29, 2026

   Owner

   More actions

   If there are no links from elsewhere I think it's probably ok to just hack the ids?
5. jhcarl0814 commented on Apr 29, 2026

   

   jhcarl0814

   on Apr 29, 2026

   ContributorAuthor

   More actions

   @timsong-cpp 👌id conflicts avoided by adding trailing underscores
6. jhcarl0814 commented on May 4, 2026

   

   jhcarl0814

   on May 4, 2026

   ContributorAuthor

   More actions

   @timsong-cpp It appears that the links of previous versions have been fixed. These are broken links in the current working draft:

   working draft, Index (https://timsong-cpp.github.io/cppwp/generalindex )

   Some terms have conflicting ids:

   contract violation, [basic.contract.eval] and contract_violation, [basic.contract.eval], [basic.contract.handler] have same id=':contract_violation'. Maybe the second item should be in "Index of library name" (https://timsong-cpp.github.io/cppwp/libraryindex )?

   Some links' #targets do not exist:

   checking semantics (https://timsong-cpp.github.io/cppwp/generalindex#:checking_semantics )'s link contract evaluation semantics\u0001checking (https://timsong-cpp.github.io/cppwp/generalindex#:contract_evaluation_semantics\u0001checking ).
   terminating semantics (https://timsong-cpp.github.io/cppwp/generalindex#:terminating_semantics )'s link contract evaluation semantics\u0001terminating (https://timsong-cpp.github.io/cppwp/generalindex#:contract_evaluation_semantics\u0001terminating ).

   ---

   working draft, Index of library names (https://timsong-cpp.github.io/cppwp/libraryindex )

   Some terms have conflicting ids:

   operator~ and operator have same id="lib:operator~". Maybe they should be merged?
7. timsong-cpp commented on May 4, 2026

   

   timsong-cpp

   on May 4, 2026

   Owner

   More actions

   @timsong-cpp It appears that the links of previous versions have been fixed. These are broken links in the current working draft:

   working draft, Index (https://timsong-cpp.github.io/cppwp/generalindex )

   Some terms have conflicting ids:

   contract violation, [basic.contract.eval] and contract_violation, [basic.contract.eval], [basic.contract.handler] have same id=':contract_violation'. Maybe the second item should be in "Index of library name" (https://timsong-cpp.github.io/cppwp/libraryindex )?

   It's going to the core language spec though.

   Some links' #targets do not exist:

   checking semantics (https://timsong-cpp.github.io/cppwp/generalindex#:checking_semantics )'s link contract evaluation semantics\u0001checking (https://timsong-cpp.github.io/cppwp/generalindex#:contract_evaluation_semantics\u0001checking ). terminating semantics (https://timsong-cpp.github.io/cppwp/generalindex#:terminating_semantics )'s link contract evaluation semantics\u0001terminating (https://timsong-cpp.github.io/cppwp/generalindex#:contract_evaluation_semantics\u0001terminating ).

   Eelis/cxxdraft-htmlgen#124

   working draft, Index of library names (https://timsong-cpp.github.io/cppwp/libraryindex )

   Some terms have conflicting ids:

   operator~ and operator have same id="lib:operator~". Maybe they should be merged?

   cplusplus/draft#9011

   Out of curiosity, what are you using to find these?
8. timsong-cpp commented on May 4, 2026

   

   timsong-cpp

   on May 4, 2026

   Owner

   More actions

   @timsong-cpp It appears that the links of previous versions have been fixed. These are broken links in the current working draft:
   working draft, Index (https://timsong-cpp.github.io/cppwp/generalindex )
   Some terms have conflicting ids:

   Looks like this is a much bigger problem and a large part of it will have to be fixed in the tool:
   Eelis/cxxdraft-htmlgen#125
9. jhcarl0814 commented on May 4, 2026

   

   jhcarl0814

   on May 4, 2026

   Last edited by jhcarl0814

   ContributorAuthor

   More actions

   Well, if those two links have to be on the same page, I hope they're close together (which is indeed the case) so that even if a link to #id redirects users to the wrong location (always the first element with id), they can quickly find the correct content next to it (other elements with id).

   Out of curiosity, what are you using to find these?

   I'm writing some programs (4 targets in CMake terms) to extract content from the site, structurize it, and use the built index to let users perform full-text search on the working drafts. This issue is raised while extracting content from the "Index" series of pages.

   At first glance, it seems that the Javascript snippets only need to build a map from terms to URLs, and when the programmer finds that the same term can appear multiple times, then build a map from terms to lists of URLs. However, there are see and see also links that link to #targets in the same page. I don't want users to be directed to those huge "Index" pages after clicking on search results. Instead, I want users searching for ! to see and click on the search results displayed when searching for logical negation, meaning that the see and see also links should be pre-followed to save users' time. So the resulting Javascript snippets should build a map from ids to indexes of element (in the Array to be returned from Javascript snippets), and the links pointing to #targets in the same page will be replaced by "index of element" by looking up in the map. When duplicate ids exist, the program can not guess which element a #target points to.

   The following Javascript snippets (to execute in the "Index" series of pages), (when I was debugging/testing) are executed by "Press F12 to open Chrome developer tools >> switch to the Console tab >> paste the snippet", or (when actually used by the web scraper) are executed by my_QWebEngineView->page()->runJavaScript(uR"js(...)js"_s) (if !result_QVariant.isValid() it means the Javascript snippet threw an Error, and I need to return to the browser and execute the Javascript snippet to see the error messages).

   For working draft (and also earlier versions; if it stops working at some version, stop using it and modify it to make it work):

   (function(){
       Array.from(document.querySelectorAll(".mjx-sub, sub")).forEach(function(e){ e.parentNode.insertBefore(document.createTextNode("_"), e); });
       Array.from(document.querySelectorAll(".mjx-sup, sup")).forEach(function(e){ e.parentNode.insertBefore(document.createTextNode("^"), e); });
       Array.from(document.querySelectorAll(".math, .mjx-chtml")).forEach(function(e){ e.replaceWith(document.createTextNode(e.textContent)); });
   
       let id_to_item_index = new Map(), url_current_page = new URL(window.location.href), items = [];
       let current_range = document.createRange();
       let assert_current_range_is_empty = function(){ window.getSelection().removeAllRanges(); window.getSelection().addRange(current_range); let s = window.getSelection().toString().trim(); if(s.length != 0) throw new Error(); };
       let encountered_error = false;
       let traverse_node = function(e, parent_index){
           current_range.setStartBefore(e); current_range.setEndBefore(e.children[0]); assert_current_range_is_empty();
   
           current_range.setStartBefore(e.children[0]);
           let end_e = e.children[0].querySelector("a:not([title]),div.indexitems");
           if(end_e.tagName == "A"){
               if(end_e.previousSibling.nodeType == Node.TEXT_NODE){
                   if(end_e.previousSibling.data.endsWith(", "))
                       current_range.setEnd(end_e.previousSibling, end_e.previousSibling.data.length - 2);
                   else if(end_e.previousSibling.data == " " && end_e.previousSibling.previousSibling.nodeType == Node.ELEMENT_NODE && end_e.previousSibling.previousSibling.tagName == "I" && (end_e.previousSibling.previousSibling.innerText == "see" || end_e.previousSibling.previousSibling.innerText == "see also") && end_e.previousSibling.previousSibling.previousSibling.data.endsWith(", "))
                       current_range.setEnd(end_e.previousSibling.previousSibling.previousSibling, end_e.previousSibling.previousSibling.previousSibling.data.length - 2);
                   else
                       throw new Error();
               }
               else
                   throw new Error();
   
           }
           else if(end_e.tagName == "DIV")
               current_range.setEndBefore(end_e);
           let item_index, term = (function(){ window.getSelection().removeAllRanges(); window.getSelection().addRange(current_range); let s = window.getSelection().toString().trim(); if(s.length != 0) return s; })().replaceAll("\u00AD", "");
           if(term.includes(", "))
               console.warn("!!! term.includes(\", \")", term);
           if(id_to_item_index.has(e.id)){
               console.warn("duplicate [id]:", e.id);
               item_index = id_to_item_index.get(e.id);
               if(items[item_index]["term"] != term)
                   encountered_error = true, console.error("duplicate [id], inconsistent term:", items[item_index]["term"], term);
           }
           else{
               item_index = items.length;
               id_to_item_index.set(e.id, item_index);
               items.push({
                   "term": term,
                   "links": [],
                   "children_index_relative": [],
                   "parent_index_relative": parent_index == -1 ? -1 : item_index - parent_index,
               });
           }
           items[item_index]["links"].push(...Array.from(e.querySelectorAll("a:not(& div[id] a)")).map(function(a){
               let url = new URL(a.href);
               if(url_current_page.protocol == url.protocol && url_current_page.username == url.username && url_current_page.password == url.password && url_current_page.hostname == url.hostname && url_current_page.port == url.port && url_current_page.pathname == url.pathname && url_current_page.search == url.search){
                   if(a.querySelector("&>b") != null)
                       throw new Error();
                   return decodeURIComponent(url.hash).substring(1);
               }
               else
                   return {
                       "anchor_href_hash": url.hash,
                       "anchor_href_without_hash": (url.hash = "", url.toString()),
                       // "anchor_text": a.innerText,
                       "is_bold": a.querySelector("&>b") != null,
                   }
           }));
           items[item_index]["children_index_relative"].push(...Array.from(e.querySelectorAll("div[id]:not(& div[id] div[id])")).map(function(e){ return traverse_node(e, item_index) - item_index; }));
           current_range.setStartAfter(e.children[0]); current_range.setEndAfter(e); assert_current_range_is_empty();
           return item_index;
       };
       Array.from(document.querySelectorAll(".wrapper>div[id]")).forEach(function(e){ traverse_node(e, -1); });
       for(let [item_index, item] of items.entries()){
           if(item["links"].length == 0 && item["children_index_relative"].length == 0)
               throw new Error();
           for(let link_index = 0; link_index != item["links"].length; ++link_index){
               if(typeof item["links"][link_index] == "string"){
                   if(!id_to_item_index.has(item["links"][link_index]))
                       encountered_error = true, console.error("!id_to_item_index.has(item[\"links\"][link_index])", item, link_index);
                   item["links"][link_index] = id_to_item_index.get(item["links"][link_index]) - item_index;
               }
           }
       }
       if(encountered_error){
           console.log(items);
           throw new Error();
       }
       return items;
   })()

   For N4861 (and also earlier versions; if it stops working at some version, stop using it and modify it to make it work):

   (function(){
       Array.from(document.querySelectorAll(".mjx-sub, sub")).forEach(function(e){ e.parentNode.insertBefore(document.createTextNode("_"), e); });
       Array.from(document.querySelectorAll(".mjx-sup, sup")).forEach(function(e){ e.parentNode.insertBefore(document.createTextNode("^"), e); });
       Array.from(document.querySelectorAll(".math, .mjx-chtml")).forEach(function(e){ e.replaceWith(document.createTextNode(e.textContent)); });
   
       let id_to_item_index = new Map(), url_current_page = new URL(window.location.href), items = [];
       let current_range = document.createRange();
       let assert_current_range_is_empty = function(){ window.getSelection().removeAllRanges(); window.getSelection().addRange(current_range); let s = window.getSelection().toString().trim(); if(s.length != 0) throw new Error(); };
       let encountered_error = false;
       let traverse_node = function(e, parent_index){
           current_range.setStartBefore(e); current_range.setEndBefore(e.children[0]); assert_current_range_is_empty();
   
           current_range.setStartBefore(e.children[0]);
           let end_e = e.children[0].querySelector("a:not(.grammarterm),div.indexitems");
           if(end_e.tagName == "A"){
               if(end_e.previousSibling.nodeType == Node.TEXT_NODE){
                   if(end_e.previousSibling.data.endsWith(", "))
                       current_range.setEnd(end_e.previousSibling, end_e.previousSibling.data.length - 2);
                   else if(end_e.previousSibling.data == " " && end_e.previousSibling.previousSibling.nodeType == Node.ELEMENT_NODE && end_e.previousSibling.previousSibling.tagName == "I" && (end_e.previousSibling.previousSibling.innerText == "see" || end_e.previousSibling.previousSibling.innerText == "see also") && end_e.previousSibling.previousSibling.previousSibling.data.endsWith(", "))
                       current_range.setEnd(end_e.previousSibling.previousSibling.previousSibling, end_e.previousSibling.previousSibling.previousSibling.data.length - 2);
                   else
                       throw new Error();
               }
               else
                   throw new Error();
   
           }
           else if(end_e.tagName == "DIV")
               current_range.setEndBefore(end_e);
           let item_index, term = (function(){ window.getSelection().removeAllRanges(); window.getSelection().addRange(current_range); let s = window.getSelection().toString().trim(); if(s.length != 0) return s; })();
           if(term.includes(", "))
               console.warn("!!! term.includes(\", \")", term);
           if(id_to_item_index.has(e.id)){
               console.warn("duplicate [id]:", e.id);
               item_index = id_to_item_index.get(e.id);
               if(items[item_index]["term"] != term)
                   encountered_error = true, console.error("duplicate [id], inconsistent term:", items[item_index]["term"], term);
           }
           else{
               item_index = items.length;
               id_to_item_index.set(e.id, item_index);
               items.push({
                   "term": term,
                   "links": [],
                   "children_index_relative": [],
                   "parent_index_relative": parent_index == -1 ? -1 : item_index - parent_index,
               });
           }
           items[item_index]["links"].push(...Array.from(e.querySelectorAll("a:not(& div[id] a)")).map(function(a){
               let url = new URL(a.href);
               if(url_current_page.protocol == url.protocol && url_current_page.username == url.username && url_current_page.password == url.password && url_current_page.hostname == url.hostname && url_current_page.port == url.port && url_current_page.pathname == url.pathname && url_current_page.search == url.search){
                   if(a.querySelector("&>b") != null)
                       throw new Error();
                   return decodeURIComponent(url.hash).substring(1);
               }
               else
                   return {
                       "anchor_href_hash": url.hash,
                       "anchor_href_without_hash": (url.hash = "", url.toString()),
                       // "anchor_text": a.innerText,
                       "is_bold": a.querySelector("&>b") != null,
                   }
           }));
           items[item_index]["children_index_relative"].push(...Array.from(e.querySelectorAll("div[id]:not(& div[id] div[id])")).map(function(e){ return traverse_node(e, item_index) - item_index; }));
           current_range.setStartAfter(e.children[0]); current_range.setEndAfter(e); assert_current_range_is_empty();
           return item_index;
       };
       Array.from(document.querySelectorAll(".wrapper>div[id]")).forEach(function(e){ traverse_node(e, -1); });
       for(let [item_index, item] of items.entries()){
           if(item["links"].length == 0 && item["children_index_relative"].length == 0)
               throw new Error();
           for(let link_index = 0; link_index != item["links"].length; ++link_index){
               if(typeof item["links"][link_index] == "string"){
                   if(!id_to_item_index.has(item["links"][link_index]))
                       encountered_error = true, console.error("!id_to_item_index.has(item[\"links\"][link_index])", item, link_index);
                   item["links"][link_index] = id_to_item_index.get(item["links"][link_index]) - item_index;
               }
           }
       }
       if(encountered_error){
           console.log(items);
           throw new Error();
       }
       return items;
   })()

   For N4659 (and also earlier versions; if it stops working at some version, stop using it and modify it to make it work):

   (function(){
       Array.from(document.querySelectorAll(".mjx-sub, sub")).forEach(function(e){ e.parentNode.insertBefore(document.createTextNode("_"), e); });
       Array.from(document.querySelectorAll(".mjx-sup, sup")).forEach(function(e){ e.parentNode.insertBefore(document.createTextNode("^"), e); });
       Array.from(document.querySelectorAll(".math, .mjx-chtml")).forEach(function(e){ e.replaceWith(document.createTextNode(e.textContent)); });
   
       let id_to_item_index = new Map(), url_current_page = new URL(window.location.href), items = [];
       let current_range = document.createRange();
       let assert_current_range_is_empty = function(){ window.getSelection().removeAllRanges(); window.getSelection().addRange(current_range); let s = window.getSelection().toString().trim(); if(s.length != 0) throw new Error(); };
       let encountered_error = false;
       let traverse_node = function(e, parent_index){
           current_range.setStartBefore(e); current_range.setEndBefore(e.children[0]); assert_current_range_is_empty();
   
           current_range.setStartBefore(e.children[0]);
           let end_e = e.children[0].querySelector("a:not(.grammarterm):not(i>a),div.indexitems");
           if(end_e.tagName == "A"){
               if(end_e.previousSibling.nodeType == Node.TEXT_NODE){
                   if(end_e.previousSibling.data.endsWith(", "))
                       current_range.setEnd(end_e.previousSibling, end_e.previousSibling.data.length - 2);
                   else if(end_e.previousSibling.data == " " && end_e.previousSibling.previousSibling.nodeType == Node.ELEMENT_NODE && end_e.previousSibling.previousSibling.tagName == "I" && (end_e.previousSibling.previousSibling.innerText == "see" || end_e.previousSibling.previousSibling.innerText == "see also") && end_e.previousSibling.previousSibling.previousSibling.data.endsWith(", "))
                       current_range.setEnd(end_e.previousSibling.previousSibling.previousSibling, end_e.previousSibling.previousSibling.previousSibling.data.length - 2);
                   else
                       throw new Error();
               }
               else
                   throw new Error();
   
           }
           else if(end_e.tagName == "DIV")
               current_range.setEndBefore(end_e);
           let item_index, term = (function(){ window.getSelection().removeAllRanges(); window.getSelection().addRange(current_range); let s = window.getSelection().toString().trim(); if(s.length != 0) return s; })().replaceAll("-\n", "");
           if(term.includes(", "))
               console.warn("!!! term.includes(\", \")", term);
           if(id_to_item_index.has(e.id)){
               console.warn("duplicate [id]:", e.id);
               item_index = id_to_item_index.get(e.id);
               if(items[item_index]["term"] != term)
                   encountered_error = true, console.error("duplicate [id], inconsistent term:", items[item_index]["term"], term);
           }
           else{
               item_index = items.length;
               id_to_item_index.set(e.id, item_index);
               items.push({
                   "term": term,
                   "links": [],
                   "children_index_relative": [],
                   "parent_index_relative": parent_index == -1 ? -1 : item_index - parent_index,
               });
           }
           items[item_index]["links"].push(...Array.from(e.querySelectorAll("a:not(& div[id] a)")).map(function(a){
               let url = new URL(a.href);
               if(url_current_page.protocol == url.protocol && url_current_page.username == url.username && url_current_page.password == url.password && url_current_page.hostname == url.hostname && url_current_page.port == url.port && url_current_page.pathname == url.pathname && url_current_page.search == url.search){
                   if(a.querySelector("&>b") != null)
                       throw new Error();
                   return decodeURIComponent(url.hash).substring(1);
               }
               else
                   return {
                       "anchor_href_hash": url.hash,
                       "anchor_href_without_hash": (url.hash = "", url.toString()),
                       // "anchor_text": a.innerText,
                       "is_bold": a.querySelector("&>b") != null,
                   }
           }));
           items[item_index]["children_index_relative"].push(...Array.from(e.querySelectorAll("div[id]:not(& div[id] div[id])")).map(function(e){ return traverse_node(e, item_index) - item_index; }));
           current_range.setStartAfter(e.children[0]); current_range.setEndAfter(e); assert_current_range_is_empty();
           return item_index;
       };
       Array.from(document.querySelectorAll(".wrapper>div[id]")).forEach(function(e){ traverse_node(e, -1); });
       for(let [item_index, item] of items.entries()){
           if(item["links"].length == 0 && item["children_index_relative"].length == 0)
               throw new Error();
           for(let link_index = 0; link_index != item["links"].length; ++link_index){
               if(typeof item["links"][link_index] == "string"){
                   if(!id_to_item_index.has(item["links"][link_index]))
                       encountered_error = true, console.error("!id_to_item_index.has(item[\"links\"][link_index])", item, link_index);
                   item["links"][link_index] = id_to_item_index.get(item["links"][link_index]) - item_index;
               }
           }
       }
       if(encountered_error){
           console.log(items);
           throw new Error();
       }
       return items;
   })()
10. jhcarl0814 commented on May 7, 2026

    

    jhcarl0814

    on May 7, 2026

    ContributorAuthor

    More actions

    @timsong-cpp The (PC-only) webpage is ready. It's incomplete because the content from https://timsong-cpp.github.io/cppwp/generalindex and https://timsong-cpp.github.io/cppwp/libraryindex are not scraped.
    https://jhcarl0814.github.io/cpp_working_drafts/cpp_working_drafts.html
    However, on the first visit, it takes about 10 seconds to download the index. The website can also be run locally by downloading the repo.

    Query parameters can be used to control settings when sharing links, e.g.:
    https://jhcarl0814.github.io/cpp_working_drafts/cpp_working_drafts.html?filter_standard_version=N3337,N4140,N4618,N4659,N4861,N4868,N4950,working_draft&filter_result_type=index_item_term,section_number,section_title,section_abbreviation,body_text_content,footnote_content&q=to_underlying