[ https://issues.apache.org/jira/browse/FLINK-18496?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17166374#comment-17166374 ]
Zhilong Hong commented on FLINK-18496: -------------------------------------- I think in fact the plugin is conflict with the excerpt rather than the {{_post}} folder. 1. After upgrading to Jekyll 4.0.1, I get errors: [https://gist.github.com/Thesharing/a4681dcef6229903088fb9eb49eaea6f] . The build stuck in endless loop when it comes to: {code:bash} /srv/flink-web/.rubydeps/ruby/2.5.0/gems/jekyll-multiple-languages-2.0.3/lib/jekyll-multiple-languages/multilang.rb:55:in `append_data_for_liquid' /srv/flink-web/.rubydeps/ruby/2.5.0/gems/jekyll-multiple-languages-2.0.3/lib/jekyll-multiple-languages/document.rb:71:in `to_liquid' {code} The function {{append_data_for_liquid}} calls {{deep_merge_hashes!}} in jekyll/utils.rb, and the implementation of {{deep_merge_hashes!}} is different between Jekyll 3 and 4. In Jekyll 3: {code:ruby} def deep_merge_hashes!(target, overwrite) overwrite.each_key do |key| if overwrite[key].is_a? Hash and target[key].is_a? Hash target[key] = Utils.deep_merge_hashes(target[key], overwrite[key]) next end target[key] = overwrite[key] end if target.default_proc.nil? target.default_proc = overwrite.default_proc end target end {code} In Jekyll 4: {code:ruby} def deep_merge_hashes!(target, overwrite) merge_values(target, overwrite) merge_default_proc(target, overwrite) duplicate_frozen_values(target) target end {code} The loop gets stuck at {{duplicate_frozen_values}}. I try to remove {{duplicate_frozen_values}} and build the doc. The build succeeds. However, the blogs without excerpts in markdown have no automatically generated excerpts in the web page. 2. I renamed the folder {{_posts}} to other names like {{articles}}. The build succeeds. But the reference of {{_posts}} in {{index.md}} is broken. Then I try to upgrade Jekyll to 4.1.1 and setting {{page_excerpts}} to {{true}}, which enables excerpts for every page. The build fails like above. So in my opinion, the issue is related to the process of generating excerpts with "jekyll-multiple-language". I'm not familiar with Ruby, and I have no idea how to debug Jekyll with IDE. I think I'm stuck here right now, and I'd really appreciate it if anyone could help me with this issue. > Anchors are not generated based on ZH characters > ------------------------------------------------ > > Key: FLINK-18496 > URL: https://issues.apache.org/jira/browse/FLINK-18496 > Project: Flink > Issue Type: Bug > Components: Project Website > Reporter: Zhu Zhu > Assignee: Zhilong Hong > Priority: Major > Labels: starter > > In ZH version pages of flink-web, the anchors are not generated based on ZH > characters. The anchor name would be like 'section-1', 'section-2' if there > is no EN characters. An example can be the links in the navigator of > https://flink.apache.org/zh/contributing/contribute-code.html > This makes it impossible to ref an anchor from the content because the anchor > name might change unexpectedly if a new section is added. > Note that it is a problem for flink-web only. The docs generated from the > flink repo can properly generate ZH anchors. -- This message was sent by Atlassian Jira (v8.3.4#803005)