Compatibility with ArXiv

[Rubiel1]

Hello,

• In my new paper, the html autogenerated page of ArXiv did not compile the \pcauset{} operator.

• I made my file in overleaf, but when I download the zip, part of the main.tex file was missing, and four times the same problem happened. The part

uset{2 ,4 ,1 ,3}$ filter is in the optimal position: inserted before the last convolution}\\*

appeared instead of

\textbf{2. Results when \pcauset{2 ,4 ,1 ,3} filter is in the optimal position: inserted before the last convolution} \\*

I think this is evidence of some incompatibility, either with ArXiv auto html system, or Overleaf submit system.

[c-minz]

Overleaf uses a full LaTeX distribution (TeX Live), which includes the causets package.
The compilation system of arXiv does not necessarily support all packages, but I have submitted two papers that use the causets package (once including the causets.sty and once without) and it seems to compile the PDF without trouble (either way). Note: to reduce compilation time of arXiv, you may use the externalisation feature of the package and included the subdirectory 'causets/' after you compiled it locally.

Currently, the new HTML feature of arXiv is in an "experimental" stage and it has not been tested with the causets package. Thank you for bringing this issue forward. To find a solution and improve the implementation of the package for HTML support, could you provide some more information about the issue.

1. Was your submission correctly processed into a PDF file on arXiv?
2. What were the steps that you did to submit?
3. At what step did some parts of the source code get lost? I understood that you drafted the paper on Overleaf, downloaded the project as a ZIP-file and then found out that some parts of the main.tex where missing. Or did you upload it to the arXiv and just afterwards noticed missing parts?
4. Did you submit the paper through the submission system of Overleaf?

[Rubiel1]

1.-Was your submission correctly processed into a PDF file on arXiv?
No, the submission was rejected, until the part

uset{2 ,4 ,1 ,3}$ filter is in the optimal position: inserted before the last convolution}\\*

was replaced by
\textbf{2. Results when \pcauset{2 ,4 ,1 ,3} filter is in the optimal position: inserted before the last convolution} \\*

After that change I was able to submit.

2.-What were the steps that you did to submit?

I compiled in overleaf.

I clicked on the submit option.

I choose the arxiv version and I downloaded my file.

I then normally submitted to arxiv my zip files.

At what step did some parts of the source code get lost? I understood that you drafted the paper on Overleaf, downloaded the project as a ZIP-file and then found out that some parts of the main.tex where missing. Or did you upload it to the arXiv and just afterwards noticed missing parts?

Historically, yes I found it after ArXiv rejected it, but I checked my download files and it turns out that the missing code was already missing on the zip files. So this seems to be an OverLeaf thing.

Did you submit the paper through the submission system of Overleaf?

No, I only download the zip file through the submission system of overleaf. ( For arxiv, Overleaf only lets you download, not submit directly as far as I know).

Note: I suspect the \pc and the \\* may have something to do.

[Rubiel1]

I also opened this issue in ArXiv arXiv/html_feedback#2766 (comment)

[Rubiel1]

Update, it was discovered that the "missing text" part was a human error. Then the only thing to report is that HTML for arxiv didnot display \pcausets propertly.

[c-minz]

Thank you for reporting, the details and update.

Indeed, I tried to reproduce your error by copying the text line into an Overleaf document and using the submission button, which did not cause any issues on my end. So this first problem is resolved.

Issue: HTML conversion on arXiv

The arXiv has a list of troubling packages, some of which seem to use TikZ as well. The diagrams of the causets package are generated with TikZ, so the HTML converter would have to be able to recognize that the LaTeX-macros create graphics, convert these into a suitable online format and replace the source macros with the graphics. I could not yet find out if the arXiv converter already supports TikZ to some degree (which would be required for a dynamic generation of diagrams). A workaround might be possible with the externalisation feature of the packages so that the diagrams have to be included in the submission (typically the automatic sub-directory created during a local compilation), and the HTML converter takes the pre-rendered PDFs and generates online graphics from these.

Diagrams are also supported in math-mode when using LaTeX. In HTML, this feature would only be possible if the math-typesetting used by arXiv supports graphics as well.