nomennescio

Run all basis tests, a quick scan seems to indicate running from Cygwin Bash or cmd.exe doesn't make a big difference [basis-test-errors.txt](https://github.com/factor/factor/files/7929464/basis-test-errors.txt)

> ```factor > { "io" } make-staging-image > ``` That triggers a bug! ``` delete-staging-images IN: scratchpad Command: continue-restart 1: Note: Added "tools.deploy.backend" vocabulary to search path IN: scratchpad {...

When looking at singleton `windows` it has methods that make explicit assumptions about the build environment (https://docs.factorcode.org/content/word-windows,system.html) ``` USING: kernel layouts mason.child sequences system ; M: windows compile-factor-command { "nmake"...

Can I help with debugging?

When I look in the listener for words generating `callsite-not-compiled` I see ``` \ callsite-not-compiled usage. Computing usage index... done IN: alien : alien-assembly ( args... return parameters abi quot...

This might be related to https://github.com/factor/factor/pull/936 https://github.com/factor/factor/issues/364 https://github.com/factor/factor/issues/495

I can understand your concern. In practice a user can redefine any Factor word anyway, so it doesn't seem like a fundamental "security" issue, but it could prevent unexpected "breakage"....

Maybe a simple compromise. might be to have a `FACTOR_WORK` environment variable, that overrides only `resource:work`? Like `FACTOR_WORK= factor` Or a command-line flag that disables it? `factor --no-work` hence `FACTOR_ROOTS=...

I guess you could consider many such things "installation" issues. If you keep the work directory of Factor which is in your path empty, and only use FACTOR_ROOTS you won't...

@mrjbq7 It seems that present and prettyprinting could work very well together. In prettyprinting for primitive types with complex formatting, it often is required to convert an object to string...