- This table has
table-layout: autoandwidth: 100% - None of the cells has a defined width
- As you can see, the width of each column is automatically adjusted to the best fit, by the browser
When I convert it to PDF with xmlworker 5.5.3 (iTextSharp 5.5.3), I get this PDF:

As you can see, iTextSharp has decided very bad the widths for the columns (compare it to how Chrome draws the table):

It is not efficient:
- Large columns are drawn too narrow (eg. 4th column)
- Short columns are drawn too wide
I tried other "html to pdf" tools and they do well with the width of the columns:
But what I like about iTextSharp is that I can use it directly on my web app hosted in Azure.
- puppeteer is made in node.js and my web app is made in C# / asp.net
- wkhtmltopdf needs to run in a B1 tier in Azure (expensive)
So... what am I doing wrong with xmlworker? or it is simply a bad library?
UPDATE 1
There is a puppeteer project for C# called "PuppeteerSharp". I'm trying now. Apparently it needs to be run along with browserless.io because Azure won't run Chrome on a web app (F1 at least...)
UPDATE 2
Not easy to run PuppeteerSharp in Azure.
For now, I deployed the GitHub project "url-to-pdf-api" in heroku, which uses Puppeteer, and I'm calling this from Azure. Not fast, not simple...
UPDATE 3
I tried updating from xmlworker 5.5.3 to xmlworker 5.5.13.2 and its even worse. It doesn't consider style="width: 25%" on <td> (which 5.5.3 does OK). What a terrible library


