Currently, colors with alpha channel defined with 'rgba' function are not supported, for example the following:

<span style="background-color: rgba(0, 0, 0, 0.5);">Some text</span>

is imported as black color; semi-transparent colors are exported without the alpha channel.

HtmlFormatProvider cannot import a document that contains an image with invalid URI format. An UriFormatException is thrown and the whole import fails. Similar errors should be handled and the document should be imported.

More information: https://developer.mozilla.org/en-US/docs/Web/HTML/Element/tt .

Note: this element is obsolete, and has the same meaning as <code>.

If span formatting command (bold, italic, strikethrough) is applied to selection over multiple paragraphs, and tracked with tracked changes; and later this tracked change is rejected, NullReferenceException is thrown. The exception is handled so the visible result for the end user is that the UI stops updating and any subsequent changes are not visible.

All Microsoft Office applications (Outlook, Exchange, Word, Excel) that can produce HTML are creating them with all of the CSS stylings commented with HTML comment, as follows: <style><!-- CSS --></style>
This is a problem as RadRichTextBox's HTML parser does not import the CSS when it is commented like that.

Workaround: Remove the HTML comment from the CSS style tag.

Currently <big> and <small> tags are not supported and the text inside is not imported.

Currently HtmlFormatProvider does not export/import tables with alignment (left, center, right).

It seems that the characters of Arabic words are reversed on export. Moreover, Arabic letters has different letter forms according to their position in the word (contextual letter forms) - beginning, end, middle, isolated. It seems that same character should be exported to PDF with different charcode depending on the character positioning. 

The invalid values should be skipped and the document should be successfully imported.

Check all possible shorthand properties, e.g. border-[left/top/right/bottom]. Also check CSS properties like border-[left/top/right/bottom]-[color/style/width].

Add support for inset borders for tables and table cells (Inset Line Border in OOXML).

Nested tables inherit alignment from its parent table or table cell when importing a document from HTML, but they shouldn't.
The paragraphs are also aligned incorrectly when their parent table has its align attribute set.

Currently ignored words are marked only in-memory using internal properties of the document model. Make them persist, similar to what MS Word do.

Note: In the DOCX format, the document is marked as "clear" (doesn't contain spell check errors) and all misspelled words are enclosed with an errProof annotation.

The word is temporary marked as correct and when another word is added to the dictionary, the ignored one is highlighted again. As a workaround could be used the Ignore All option.

When adding a comment it should be marked as created from the current user. Layer with tool tips should be added to show this information. Also comment balloons should contain it.

Improve the performance when multiple paragraphs are selected and deleted.

When trying to import a document containing bookmark and BookmarkRangeEnd is after the last paragraph, a StackOverflowException is thrown in IntervalNode while trying to rebuild the document cache. (RadDocumentChildrenCache.RebuildTree()).

The same could be observed if the bookmark end is last in the header/footer, between table cells in the RadDocument body, or when all paragraphs after the bookmark range end are hidden (with vanish property).

Workaround: Detect the problematic annotations and remove them:
foreach (var rangeStart in document.EnumerateChildrenOfType<AnnotationRangeStart>())
{
    if (rangeStart.End == null || rangeStart.End.Parent == null || rangeStart.End.Parent.Parent == null)
    {
        //MessageBox.Show("This indicates a problem.");
        rangeStart.Parent.Children.Remove(rangeStart);
    }
}

Note: Until 2017 R2 SP1 release StackOverflowException is thrown, after this release an ArgumentException should be thrown.

When the specified length of the stream is larger than the actual one, Adobe throws errors and removes the image.

Workaround:

Set the format provider settings:

PdfFormatProvider provider = new PdfFormatProvider();
PdfExportSettings exportSettings = new PdfExportSettings();
exportSettings.ContentsDeflaterCompressionLevel = 9;
exportSettings.ImagesDeflaterCompressionLevel = 9;
provider.ExportSettings = exportSettings;

When the paragraph formatting symbol has a formatting, different than the one of the spans in this paragraph, the bullet is not aligned to the baseline of the text