The -source command now runs with a temporary namespace that is amalgamated
from the local and up namespace of the caller -source; this means that it can
no longer mutate its caller's local scope, which is the only possible sensible
behavior anyway.
This fixes#1202.
Most of the places that need to directly call a function is in the edit package,
which need to call user-defined callbacks.
This change eliminates most call sites of NewTopFrame (including all call sites
outside the eval package). Remove the function and inline it in the remaining
few call sites.
Remove NewTopFrame means that the eval package no longer offers other packages
a way to construct Frame instances. This is intended: Frame is a relatively
low-level concept, and all code outside the eval package now uses the more
high-level Eval, Call, Check/CheckTree methods of *Evaler. The most notable
exception is packages that implement modules; they may still use Frame to access
the information kept in it, but they never construct Frame instances.
In future, the Frame type can be changed to an interface.
This feature supersedes the CompileWithGlobal method, and simplifies the only
use of that method in pkg/edit, where the default binding is evaluated using the
edit: namespace as the global Ns.
The Compile method was used only by the syntax highlighting code to find
compilation errors. Since it only needs the error part but not the Op part,
provide an alternative API that only exposes the error.
The new (*Evaler).Check method allows the implementation to be simpler.
Also test the JSON format feature in an integration test instead of unit test;
this makes it unnecessary for the eval package to expose NewCompilationError.
This method has the property that it always tries to compile the code even if
there is a parse failure. This is the more desirable behavior when checking
code: if there is a parse failure near the end of a chunk of code, the user may
like to learn about compile errors earlier in the code.
The default behavior of `go test -cover` ignores coverage by integration
tests. That is, it ignores coverage by tests in a different package. This
penalizes Elvish since Elvish has a lot of integration tests that are
meant to indirectly, but explicitly, exercise code in other packages that
do not have explicit unit tests.
This change makes the test coverage more accurate for the purposes of
this project.
Resolves#1062Resolves#1187
This avoids the use of the more low-level eval.NewTopFrame call, and is another
step in making (*Evaler).Eval a unified entry point for evaluating code.
The test contained a race condition: when the mock TimeAfter implementation
sends the process a SIGINT, the "sleep" function may not have entered the
"select" block, meaning that the signal will be ignored.
This commit works around this by waiting 1ms (scaled with
$ELVISH_TEST_TIME_SCALE) before sending SIGINT. The test now consistently
succeeds on my local laptop.
Introduces two functions, PipePort and CapturePort, and implement output capture
in terms of them. These two functions return *Port instances, which can also be
used in (*Evaler).Eval calls.
I noticed that most places in the documentation that emit a "note"
about Elvish behavior use `**Note**:`. There were two places that use
all uppercase. Make those consistent with the other uses. Also clarify
the note associated with the `not` builtin.
Including the float64 data type in the prompt value stream causes an error.
This causes those values to be implicitly converted to a string as happens
everywhere else in Elvish.
I initially intended to modify the code to do the implicit string coercion
for any type not otherwise explicitly handled by the pkg/ui `Concat`
methods. I decided against that approach because doing so doesn't make
sense for some types that might appear in the value stream; e.g., Elvish
exceptions.
Fixes#1186
Commit 734eb95 changed most uses of `fm.ports[x].` to a public method
that makes the intent clearer. This changes a couple of uses that were
overlooked by that prior change.
As with commit #eb2a792 I acknowledge that `golint` recommendations are
controversial and should not automatically be acted on. Nonetheless,
this change fixes legitimate lint issues such as copy/paste cleanups,
style problems I introduced (`i += 1` versus `i++`) or method/var comments
that are not in the preferred form.
It should be possible to bind a key to a literal control character. For
example, it should be possible to write:
edit:insert:binding["\e"] = $edit:command:start~
as a logically equivalent alternative to the preferred form:
edit:insert:binding[Ctrl-'['] = $edit:command:start~
Resolves#1185
Document the format key of names and modifiers; albeit with a TODO
regarding the `Shift` modifier.
Make function key modifier matching case sensitive (`Alt` not `alt`)
to match key name matching (`Enter` not `enter`).
Implement `Ctrl-?` as the logical counterpart to `"\c?"`.
Resolves#1163
There are some Navigation mode corner cases whose current behavior is
surprising. Such as inserting a space before the filename when the space
isn't wanted or redundant. Also, selecting an empty directory and pressing
Enter should insert nothing rather than an empty string.
Fixes#1169
Not sure what happened but my fix for issue #1120 resulted in the core
unit test module not being executed due to a bogus "// +build wtf"
comment. Apparently I was experimenting with "+build" directives and
forgot to remove that line. This also improves the test coverage of the
$pwd code from 44% to 100%.
