[MlMt] Can't reply without HTML/Markdown spewage???

[Bill Cole]

On 15 Aug 2017, at 9:16, Benny Kjær Nielsen wrote:

> On 14 Aug 2017, at 16:44, Bill Cole wrote:
>
>> I'm pretty sure this is new-ish behavior, since I know it has not 
>> always been this way, although I guess it might be something I 
>> changed at some point. I am running r6090 on El Capitan.
>
> I don't think it's something you have changed. Although I cannot 
> recall the details then I probably just improved the HTML to text 
> converter at some point.
>
>>> On behalf of the team at ZipRecruiter - welcome! We're experimenting 
>>> with a new matching algorithm and the search engine [found a 
>>> job](<https://u2321971.ct.sendgrid.net/wf/click?{1823 random-ish 
>>> characters}>) that might fit what you're looking for. Check out the 
>>> description and hit ["1-Click 
>>> Apply"](<https://u2321971.ct.sendgrid.net/wf/click?{1794 random-ish 
>>> characters}>) if you like what you see.
>>
>> This is ridiculously unreadable, and does not even reflect the 
>> original HTML, as the [text](<url>) structure is apparently a 
>> Markdown thing.
>
> Note that Markdown is supposed to be simple readable plain text and 
> therefore it's also reasonable to convert to Markdown when trying to 
> generate plain text from HTML. For example, this should result in 
> reasonably good conversions of emphasized text or HTML lists. In 
> general, MailMate always tries to loose as little information as 
> possible.

Yes, but when converting to "plain text" it is disconcerting to have 
body text in the middle of a sentence wrapped in brackets and followed 
by an URL of any size. I don't generate HTML mail by choice, so using 
Markdown's *obtrusive* features in generating a reply just makes it less 
comprehensible. (But I'm a Luddite who believes lightweight markup 
reached its peak expression ~25ya with setext...)

> BUT these ridiculously long URLs are certainly a problem, especially 
> when using the plain text non-Markdown mode of MailMate. I guess 
> MailMate could automatically strip (or “not create”) the Markdown 
> URLs when the user does not use Markdown by default, but I think it 
> should be optional. Maybe it should even be an integer option only 
> stripping links longer than X characters. You could then set it to 0.
>
> Does that sound like an acceptable solution for you?

Yes.

Since you're using the Python html2text, I think there the options that 
make links into per-paragraph or per-document footnotes might be a 
better general solution in addition to supporting no link encoding at 
all.

>> I don't want or need the original URLs in any reply. I don't want or 
>> need Markdown. I absolutely do not need or want a blob of "quoted" 
>> text which is neither the text I saw in the original nor the HTML of 
>> the original. I may be misremembering but I believe that MM used to 
>> do what I want it to do. How can I get that back?
>
> A quick fix is to find the file within the application bundle which 
> uses this command:
>
> 	"${MM_BUNDLE_SUPPORT}/html2text"
>
> And then add `--ignore-links` to its arguments, but this change is of 
> course overwritten whenever you update MailMate.

Thanks. Instead, I changed "IGNORE_ANCHORS = False" to "IGNORE_ANCHORS = 
True" in the relevant config.py and removed config.pyc to assure it 
would be rebuilt. I assume that will still be reverted by the next 
update, but it is also a bit more universal, applying to all places 
where html2text is used.