[tex4ht] [bug #618] Incomplete XML Document, domfilter error, truncated build on large file.

[Nasser M. Abbasi]

On 12/15/2023 6:21 AM, CV Radhakrishnan wrote:
> On 12/14/23 20:32, Nasser M. Abbasi wrote:
> 
> [...]
> 
>> The above build.sh assumes you are using TL 2023, if not please change
>> the path shown.
> 
> Yes, I have TL 2023.
> 
>> After you start ./build.sh it will take about 12-16 hrs or so depending
>> on the speed and memory of your PC.
> 
> This is the biggest problem. In the first place, the single source TeX
> file with huge size (145MB) makes it an unmanageable beast. Need to
> waste another 12 to 18h of machine time even if we could pinpoint the
> error causing line.
> 
> Hence, I split the source into several files of each with 100,000 lines.
> Located that the table in line number 382889 (original source) had

Yes. Exactly ! This is where the problem is. It is at that
table, where tex4ht stops and gives the XML and domfilter error
and truncate everything.

And yes, splitting the document to smaller files fixes the problem
and this is what I ended doing. I split the file to 4 files.

> broken the html generation which I found its place in the fourth segment
> of split files (file-4.tex).
> 
> When I ran this by inputting to the introductory chapter to replicate
> the problem, it went well to my surprise. Of course, it is not usable as
> the cross references would be wrong.
> 
> I had encountered two problems.
> 
>    * Found two packages missing (i) myspace.sty and (ii) maple.sty. Would
>      you kindly send these as well?

Sorry. I've updated the ZIP file with these missing files.

<https://12000.org/tmp/tex4ht_XML_DVI_problem/210_Hebisch.zip>

I've also put them there outside the zip file in case you just need
to download them on their own without having to download the whole
zip file

<https://12000.org/tmp/tex4ht_XML_DVI_problem/maple.sty>
<https://12000.org/tmp/tex4ht_XML_DVI_problem/myspaces.4ht>
<https://12000.org/tmp/tex4ht_XML_DVI_problem/myspaces.sty>


>    * \usepackage[]{marginnote} has been loaded without the option 'quiet'
>      that caused unwanted verbose aux writing about the xpos of the
>      margin notes which should have been handled in a different way in
>      html mode (with css directive like float) so that TeX will not spend
>      too much time on the computation of the exact location of the margin
>      note in the dvi. That indeed sped up my part compilation.
> 

Sure. I will add that option on my end. I did not know marginnote has quite mode.

> The document in whole is running in another system, will get back with
> the results soon.
> 
> Regards

Thanks for the help and for trying it.

--Nasser