#7889 Improve markdown logic for cached vs threshold limits
The markdown_render_max_length param is an absolute limit to how big something can be for markdown conversion. But cached_convert() was added later and is a better approach. E.g. a very large ticket only has to get converted once and then its good, so a markdown_render_max_length isn't needed (we even have it set very high in development.ini because of this).
However, not all markdown conversions go through cached_convert(), for example, commit messages, outbound email bodies, or markdown files in repos.
So we should have markdown_render_max_length apply only to non-cached conversions and change its default value to something not so extremely high, maybe 100000. Or possibly introduce a second option, one for cached and one for non-cached conversions. Or develop a new caching mechanism that works for markdown text that isn't part of an artifact.
An example I came across was thousands of lines like this in the commit message itself, totaling around 500k chars:
renamed: ts/5.1/fastboot/fastboot-mangle -> ts/5.2/fastboot/fastboot-mangle renamed: ts/5.1/fastboot/lib-boot -> ts/5.2/fastboot/lib-boot renamed: ts/5.1/fastboot/lib-rw -> ts/5.2/fastboot/lib-rw renamed: ts/5.1/fastboot/usr-rw -> ts/5.2/fastboot/usr-rw renamed: ts/5.1/kernel/alias/acpi -> ts/5.2/kernel/alias/acpi renamed: ts/5.1/kernel/alias/pcm -> ts/5.2/kernel/alias/pcm
Lowered the overall limit from 999999999 to 100k. Added a parameter to
ForgeMarkdown.convertto bypass the char limit.QA at hs/7889