[dvipdfmx] Intermittent xdvipdfmx failure

[Shunsaku Hirata]

Dou itashimashite, Taylor-san.

I reminded that the environment variable SOURCE_DATE_EPOCH can be used
for this purpose.
The following two different runs should generate identical text files
test-1.txt and test-2.txt

SOURCE_DATE_EPOCH=1456304492 dvipdfmx -z 0 test.xdv && hexdump
test.pdf > test-1.txt
SOURCE_DATE_EPOCH=1456304492 dvipdfmx -z 0 test.xdv && hexdump
test.pdf > test-2.txt

Thanks,
Shunsaku Hirata

2021年1月25日(月) 22:31 Philip Taylor <P.Taylor at hellenic-institute.uk>:
>
> Arigato gozaimasu, Hirata-san.  I seem to recall that someone (perhaps David Carlisle, or another member of the LaTeX team) asked for a way to generate 100% identical PDF files for the purposes of regression testing, so I will search back in the archives to see if I can locate the thread and then add the necessary additional command-line qualifier(s) required to accomplish this.
>
> Philip Taylor
> --------
> Shunsaku Hirata wrote:
>
> Thanks for sending me the files.
>
> I verified the provided files and found that there are no problems.
> The only difference other than that I mentioned before is how ID
> strings are encoded: hex strings or literal PDF strings.
> (Sorry I hadn't noticed that such differences can occur)
>
> I don't remember it very well but there is a way to generate the
> same IDs for all generated PDF files. By doing so the outputs
> should be of the same number of bytes.
>
> Thanks,
> Shunsaku Hirata
>
>