It seems comment #6 was implemented. The original problem is fixed. Closing.

No. Now we assume that the user has to insert * before <mathd>. So, this will become integrate(f3*1,x).

What if the user wanted to integrate a function named "f3"? wouldn't that be expanded into "integrate(f31,x)" ?

I have a good idea. Let's rewrite "<mathd>" to "1,". Then "<bigint><something>*<mathd>x" will become "integrate(<something>*1,x)". Also, "<bigint><mathd>x" will become "integrate(1,x)" (currently, it produces "integrate(,x)"  a syntax error).
If there will be no objections, I'll commit this change (to reduce plugin, too).

It would be nice if x*d x could be made to work, if this is not too difficult (maybe simply rewrite <mathd> into <mathd> or so). Mathematically speaking, this is the correct semantics (xd x being the differential form to be integrated).

Understood.
This requirement is however not stated in the documentation for the plugin. Maybe the whole arXiv document could be added as another subsection of the plugin's docs.

Users of the maxima plugin are supposed to read arXiv:cs/0504039. It is clearly stated there that there should be no * before <mathd>.

I think dx should be preceded by "apply", multiplying doesn't make any sense to me. But it could to other people, so indeed, how would the user know?
There currently is a rule which replaces <mathd> with a comma. Maybe we could add a new one also replacing *<mathd>? I tried modifying maximainput.scm but had no luck with a naive attempt. Anyway, I'm not sure this is a good idea.
Andrey, any comments?

In the present version of texmacs (1.0.7.15) the missing parenthesis problem reported here has been fixed, so we could close the bug.
However, the sample equation in the document still does not work because of the (invisible) * before <mathd>. Maxima then complains :

(%i1) tmint(a,b,(x^2+3*x)*,x)
incorrect syntax: , is not a prefix operator
mint(a,b,(x^2+3*x)*,
^

(note the missing "t" in the output)
For the evaluation to proceed (and obtain the expected result) one needs to replace this hidden multiplication by "apply" (space) or nothing.
So the question now is: should dx be entered with * or apply? And how would the user know what is expected?

Debian user Karl Hegbloom reports the following error :
The integral form: I_a^b EXPR * mathd x
(SF5 I _ a right ^ b right EXPR * d tab tab x)
.... when highlit in a document, copied (Escw), and pasted into a
Maxima session, translates to "int(a,b,EXPR)*mathdx", and this is
incorrect. It should look like: "integrate(EXPR,x,a,b)" instead.
When I asked Karl he confirmed that he is indeed pasting
the cut text into a maxima session running inside texmacs.
I asked Karl to send me a texmacs document producing this
error, it is attached.
