Because it's only common decency to give people the option to remove creds.

Some ideas to help google provide proper hashtag urls when searching for an article.

1. Use html5 <article> tags instead of <div> tags for the entries.
   Entry.jsx line 42 would simply become <article className="entry" id={id}> and the closing tag should match.

2. Add metadata to the entries in Entry.jsx as follows (line 39 onwards):

       var posted = (new Date(this.state.published)).toLocaleString();
       var updated = (new Date(this.state.updated)).toLocaleString();
       var postedISO = (new Date(this.state.published)).toISOString();
       var updatedISO = (new Date(this.state.updated)).toISOString();
       return (
           <div className="entry" id={id} itemscope itemtype="http://schema.org/BlogPosting">
             {deletebutton}
             <header>
               <h1 itemprop="name"><a href={idLink} itemprop="url">{this.state.title}</a></h1>
               <h2>Originally posted on <time datetime="{postedISO}" itemprop="datePublished">{posted}</time>, 
                 last updated on <time datetime="{postedISO}" itemprop="dateModified">{updated}</time>
               </h2>
             </header>
             <MarkDown ref="markdown" hidden={this.state.editing} text={this.state.postdata} onClick={this.edit} />
             <Editor ref="editor" hidden={!this.state.editing} text={text} update={this.update} view={this.view} delete={this.delete} />
             <a className="comments" href={this.props.issues}>leave a comment on github</a>
             <Tags disabled={!this.props.editable} tags={this.state.tags} onChange={this.updateTags}/>
           </div>
   

   You could also add itemprop="articleBody" to the <div> in Markdown.jsx

At a guess the first solution might be sufficient, but the second could be so much better.

Google provide a structured data tester at https://developers.google.com/structured-data/testing-tool/
I have tested one of the articles from your blog with those properties inserted at those positions in the generated html. The results are pretty nice, if I say so myself.

I don't think it matters that the url listed in the first row is wrong, since the url is correctly found later on. I could be wrong, but I don't see any means of testing it other than trying it out on a live site and seeing what happens.

Pull requests seem like so much hassle sometimes, so feel free to use the above code yourself.

add in the image dropping that pomax.github.io currently supports

i.e. on-page CSS changing with live updates, and a push into github when the CSS is acceptable.

e.g. MathJax after markdown processing, etc.

This can be improved. If there's a math processor like MathJax active, for instance, this makes typing super slow.

Let users set a custom published time for a post, and skip over any entry with a published value indicating some point in the future, also leaving them out of the RSS feed

right now the toc requires an offline regeneration. Not sure useful.

post metadata should be loaded, full stop, even if posts themselves aren't. This might require the metadata to become a single file...

Right now this information is console logs. This should be some kind of visual signal to the user in the browser, instead

if text, paste text. If image data, turn into blob, ask for filename, upload to github as image, then "paste" the image link into the post.

current data is using date/time stamps based on localeString rather than UTC

Allow users to make a post as draft so that it never shows up for unauthenticated users and doesn't end up in the RSS feed

If I understand this right, the entire blog is hidden behind index.html which uses javascript to show content. Thus a robot that doesn't run javascript won't see anything.

Is that an issue? Does Google manage to reference your blog anyway? My guess would be yes, but you might get slightly downranked because of the javascript dependancy.

Could the code be altered to write html files for each post and update index.html?

switch over to https://github.com/michael/github for the authorised client-side work

Probably best as <WebLog title="..." description="..." base="..." feeds="tag1, tag2, ..."> so that users can simply set this up in the HTML include and then never worry about it again.

dedicated #gh-weblog-.... posts can do with a next/previous link to go through the list of known blog entries

We don't want user.github.io and user.github.io/somerepo/ to clobber each other's administrative settings.

probably also requires rewriting all the stuff to CommonJS format so that Browserify does the right thing:

gulp tasks

• turn the React JSX into .js.
• concat the mixins in
• concat the dependencies in
• uglify the resulting browser package.
• make the final thing an easy to copy/paste snippet for putting on your website.