Preamble support breaks (Theorem environment is broken)

[rspuzio]

If one types

\newtheorem{defn}{Definition}

in the preamble and

\begin{defn} Definition \end{defn}

in the body, it should typeset a definition. Instead, it writes a big red "\defthm" at the top of the output and a "\defn" where the definition should have been. For an example, please see the test object:

http://planetmath.org/testobj

[holtzermann17]

Thanks Ray, that sounds like a LaTeXML configuration issue. @dginev what do you advise?

[dginev]

Compiling this file locally works well, both with pdflatex and latexml. However, is there something being sent to LaTeXML besides that file? The file attempts to be self contained, but if another preamble is set additionally things might break badly.

Indeed, pasting the source at http://latexml.mathweb.org/editor works as expected. You probably need to not send an additional preamble though.

[dginev]

OR, am I missing the separation here? I am looking at the source tab at: http://planetmath.org/node/42485/source Is that what I should be looking at?

[holtzermann17]

http://latexml.planetmath.org/editor also gives good results, using the "Metadata" profile and pasting in the source code. And I've just gotten the article to build without an error after adding a % to the end of body.

What might be going on: maybe @rspuzio has a default preamble that doesn't include amsthm...? No, I tried switching users and as rspuzio I added another % and the article is still working.

It seems like a transient bug. Can we reproduce in another article...?

[rspuzio]

It definitely looks like there is something flaky around here. I saw that, after Joe's fixes, everything looked good. To test it further, I added a theorem environment and a theorem. It checked out fine with the LaTeXML web service, but on PM it is a worse disaster --- now it claims not to understand any definitions made in the preamble, including foo and goo, which have nothing to do with amsthm.

[holtzermann17]

I edited the article again (logged in as you) and fixed it again. So, maybe we're starting to narrow this down to a... ¿browser specific issue? What browser are you using?

[dginev]

This can't be browser specific, it's LaTeXML failing due to the input. My only gues is that the preamble and the article body aren't playing well together (or what is sent to LaTeXML ends up invalid somehow).

[holtzermann17]

My thought was that Ray's browser is assembling things wrong (e.g. there's a little Javascript involved with the way preambles are displayed, maybe that's the issue?). The thing is - I sent the (")same(") preamble and body through Planetary, logged in as the same user, and got back a correctly rendered result.

But of course ultimately I agree with you Deyan, that whatever is sent is producing a LaTeXML error.

@rspuzio One thing we could do is set up a time to do some "live debugging" - you can press save again and trigger the error, and I can see what's happening in the backend, so we can at least see what's getting assembled and sent.

[rspuzio]

One thing we could do is set up a time to do some "live debugging"

Let's do that this Saturday.

[holtzermann17]

@rspuzio and I tracked this down to a problem with previewing, e.g. as described in #281 (which I haven't fixed yet). Basically, preambles aren't getting handled right in the preview, and then if you save what you previewed, you get the red ink...

Since I almost never preview anything, that's why I hadn't seen this bug before. Now that it's traced properly, it should be relatively easy to fix.