agude / agude.github.io Goto Github PK

Use {{ page.path }} to link to the history of posts on github.

Could also use a "Has been updated" flag to turn on only for certain posts.

https://developers.google.com/search/docs/data-types/article#non-amp

See the non-AMP information, especially around time of modification, etc.

Look at Ben's blog post: https://ben.balter.com/2014/03/13/pages-anchor-links/

https://www.alexgude.com should redirect to https://alexgude.com. It does not. Instead it gives a "site can not be reached error".

The desired behavior is apparently not yet supported for https domains, see: https://github.community/t5/GitHub-Pages/Does-GitHub-Pages-Support-HTTPS-for-www-and-subdomains/td-p/7116

The code implementing the card class has a bug. I set self._value but do not use it for comparison.

See http://news.ycombinator.com/item?id=21119553

This will help prevent cross-site scripting: https://developer.mozilla.org/en-US/docs/Web/HTTP/CSP

The prefered way to get updates is RSS, so it should be obvious that this is possible!

See https://www.w3.org/TR/aria-in-html/#presentation

Hi, Alex!

I was reading your cool post about animation bliting, you mentioned there:

Fortunately there is a significantly faster alternative: matplotlib’s animation blitting. Blitting increased rendering speed by a factor of 20!

But I don't see you are actually setting blit=True in FuncAnimation. This is from your notebook:

# Generate the animation
anim = animation.FuncAnimation(
    fig=fig,
    func=update,
    frames=step,
    init_func=init,
    save_count=150, #len(list(step())),
    repeat_delay=5000,
)

and by default bliting is not used:

    blit : bool, default: False
        Whether blitting is used to optimize drawing.  Note: when using
        blitting, any animated artists will be drawn according to their zorder;
        however, they will be drawn on top of any previous artists, regardless
        of their zorder.

I also read matplotlib implementation of Animation.save method:

for data in zip(*[a.new_saved_frame_seq() for a in all_anim]):
    for anim, d in zip(all_anim, data):
        # TODO: See if turning off blit is really necessary
        anim._draw_next_frame(d, blit=False)
        if progress_callback is not None:
            progress_callback(frame_number, total_frames)
            frame_number += 1
    writer.grab_frame(**savefig_kwargs)

and it looks that blit just set to False there. It probably works in live/interactive mode, but not when saving to file.

So I really wonder how did you manage to get such performance gain?

They should show up like the author pages in their own section.

Some bugs were introduced in #35. Specifically:

• Footnotes are now misaligned.
• There are dashes from <li> in the social sharing buttons!

Thanks for the great exercise Alex! i.e. https://github.com/agude/agude.github.io/blob/master/files/interview-prep/Interview%20Prep%20Python.ipynb. It's indeed very RAM-consuming, my computer can't handle the whole db but a small portion of it. This is not really an issue but when we randomly sub sample collisions and parties, many of the case_id don't match, thus a lot of rows in one table is not relevant to that in the other. I wonder if it's better to randomly sample one table, and use case_id to select the other...

If I do not leave blank lines at the top of article_card.html, then the rendering breaks.

Instead of generating a page for each series, just make the author page, and then link to the subsection. The naming is deterministic, we could just do that.

So The Book of the New Sun should link to /books/authors/gene_wolfe/#the-book-of-the-new-sun.

There are several posts that should link to my negotiating post (PR #32).

Here is an incomplete list:

• The 2020 interviewing post
• The salary posts

https://dockyard.com/blog/2019/05/03/just-details-no-devil

On this post: https://github.com/agude/agude.github.io/blob/master/_posts/2020-12-14-data_science_interview_prep_data_manipulation.md

If an article is in a category that only has one or two other articles, and one of those articles is one of the most recent articles, then there will be duplicate entries in the Recent Article section.

This can be fixed by keeping track of the articles added already.

Adjust the padding/margin of the various elements of the page (footnotes, sharing buttons, etc.) to be more consistent in the use of whitespace.

https://internetingishard.com/html-and-css/ Will be helpful.

Consider some of the advice here when setting font sizes: https://alistapart.com/article/more-meaningful-typography/

Sometimes my notebooks aren't URL encoded, so the links look like this: Notebook Thing.ipynb

https://github.com/agude/agude.github.io/blob/master/files/interview-prep/Interview%20Prep%20Python.ipynb

That pages takes almost 10 seconds to load! It has a large header image, but also the SVG plots have thousands of data points which probably slows down rendering.

Consider switching them to PNGs!

https://css-tricks.com/really-dislike-fout-font-display-optional-might-jam/

This is a "post" type item (with related links on the bottom), but should not be: http://localhost:4000/blog/cleverly-worst-code/the-code-itself/

The updated article text ("This article was updated!") should be a template to abstract away the syntax.

Be mindful of: jekyll/jekyll#1728.

See, for example, here: https://alistapart.com/article/web-typography-tables/

On small screens, the elements of the sidebar overlap.

When an article is updated, add a date to the footer note.