Generate Debug for sys types by GuillaumeGomez · Pull Request #488 · gtk-rs/gir

GuillaumeGomez · 2017-11-10T15:24:38Z

Fixes #486.

GuillaumeGomez · 2017-11-10T17:22:41Z

All green. :)

EPashkin · 2017-11-10T18:03:38Z

            try!(writeln!(
                w,
-                "{comment}#[repr(C)]\n{comment}pub struct {name} {{",
+                "{comment}#[repr(C)]\n{debug}{comment}pub struct {name} {{",


This line seems generated struct with extra line:

#[repr(C)] pub struct AtkAttribute { pub name: *mut c_char, pub value: *mut c_char, }

Wo. Where does this backline comes from? Oo

EPashkin · 2017-11-10T18:05:44Z

+                            {fields_name})\n\
+                        \t}}\n\
+                    }}\n",
+                    name = klass.c_type,


Sorry, but this unreadable, and seems repeated at minimum twice.

True. I'll write a function and maybe comment or create temporary variables in order to make this less horrible.

GuillaumeGomez · 2017-11-10T18:06:12Z

            try!(writeln!(
                w,
-                "{}#[repr(C)]\n{0}pub struct {} {{",
+                "{}#[repr(C)]\n{}\n{0}pub struct {} {{",


@EPashkin: It comes from here actually.

EPashkin · 2017-11-10T18:09:33Z

+                        \t\twrite!(f, \"{name} @ {{:?}}\", self as *const _)\n\
+                    \t}}\n\
+                }}\n",
+                name=record.c_type));


IMHO this better extract as function too and maybe use raw string for format

EPashkin · 2017-11-10T18:13:59Z

This intended that inner fields not comma-separated?

impl ::std::fmt::Debug for AtkObject {
    fn fmt(&self, f: &mut ::std::fmt::Formatter) -> ::std::fmt::Result {
        write!(f, "AtkObject @ {:?} {{ {:?} {:?} {:?} {:?} {:?} {:?} {:?} }}",
               self as *const _,
               self.parent,
               self.description,
               self.name,
               self.accessible_parent,
               self.role,
               self.relation_set,
               self.layer)
    }
}

GuillaumeGomez · 2017-11-10T18:15:45Z

Hum yes. Do you want comma?

EPashkin · 2017-11-10T18:18:21Z

Then its normal, just strange.

GuillaumeGomez · 2017-11-11T00:04:52Z

I'll add the comma tomorrow then. ;)

GuillaumeGomez · 2017-11-11T11:10:40Z

Updated.

sdroege · 2017-11-11T11:29:47Z

For printing things struct-like, please use the functions from Formatter like debug_struct()

EPashkin · 2017-11-11T11:30:41Z

+        w,
+        name,
+        &format!(
+            "write!(f, \"{name} @ {{:?}} {{{{ {fields}}}}}\",\n\


IMHO extra space needed {{{{ {fields} }}}}

EPashkin · 2017-11-11T11:31:15Z

+        impl_content)
+}
+
+fn generate_fields_with_debug(w: &mut Write, name: &str, lines: &[String]) -> Result<()> {


Not generate_debug_with_fields?

GuillaumeGomez · 2017-11-11T12:27:33Z

Updated and improved. Thanks for the suggestion @sdroege!

EPashkin · 2017-11-11T12:32:02Z

+                 } else {
+                    let field_name = field.split(":").next().unwrap().trim().replace("pub ", "");
+                    Some(if field.replace(",", "").trim().ends_with("c_void") {
+                             format!("\t\t .field(\"{name}\", &\"c_void\")\n", name=field_name)


Is really need write constant string in debug result?

I meant: maybe better just skip this field?
Current results:

//gobject impl ::std::fmt::Debug for GParamSpecString { fn fmt(&self, f: &mut ::std::fmt::Formatter) -> ::std::fmt::Result { f.debug_struct(&format!("GParamSpecString @ {:?}", self as *const _)) .field("parent_instance", &self.parent_instance) .field("default_value", &self.default_value) .field("cset_first", &self.cset_first) .field("cset_nth", &self.cset_nth) .field("substitutor", &self.substitutor) .field("_truncated_record_marker", &"c_void") .finish() } } //gtk impl ::std::fmt::Debug for GtkRcStyle { fn fmt(&self, f: &mut ::std::fmt::Formatter) -> ::std::fmt::Result { f.debug_struct(&format!("GtkRcStyle @ {:?}", self as *const _)) .field("parent_instance", &self.parent_instance) .field("name", &self.name) .field("bg_pixmap_name", &self.bg_pixmap_name) .field("font_desc", &self.font_desc) .field("color_flags", &self.color_flags) .field("fg", &self.fg) .field("bg", &self.bg) .field("text", &self.text) .field("base", &self.base) .field("xthickness", &self.xthickness) .field("ythickness", &self.ythickness) .field("rc_properties", &self.rc_properties) .field("rc_style_lists", &self.rc_style_lists) .field("icon_factories", &self.icon_factories) .field("_truncated_record_marker", &"c_void") .finish() } }

I want all fields to be there. So better having a useless "c_void" string than nothing from my point of view.

EPashkin · 2017-11-11T12:37:40Z

@sdroege, is good for you now?

sdroege · 2017-11-11T12:41:41Z

+                w,
+                &klass.c_type,
+                &format!("write!(f, \"{name} @ {{:?}}\", self as *const _)",
+                         name=klass.c_type)


You could use debug_tuple(name).field(the_pointer_string).build() here

sdroege · 2017-11-11T12:42:02Z

+            w,
+            &interface.c_type,
+            &format!("write!(f, \"{name} @ {{:?}}\", self as *const _)",
+                     name=interface.c_type)


sdroege · 2017-11-11T12:42:18Z

+                w,
+                &record.c_type,
+                &format!("write!(f, \"{name} @ {{:?}}\", self as *const _)",
+                         name=record.c_type)


sdroege · 2017-11-11T12:44:17Z

Feel free to ignore that comment, it's all good to go for me.

GuillaumeGomez · 2017-11-11T12:47:00Z

To give @EPashkin a bit of context: @sdroege and I discussed about it and globally what came out was:

me: I don't see the use for the debug_tuple. If we have the need of it some day, I'll add it but for now it's not really useful.
@sdroege: Ok.

GuillaumeGomez · 2017-11-11T14:38:05Z

I'll let you handle the merging @EPashkin.

GuillaumeGomez · 2017-11-11T15:29:36Z

Sorry but I want to quickly move towards the next release so I'll merge this and make a PR for the sys repository.

EPashkin · 2017-11-11T16:48:09Z

Thanks, @GuillaumeGomez

EPashkin reviewed Nov 10, 2017

View reviewed changes

GuillaumeGomez commented Nov 10, 2017

View reviewed changes

GuillaumeGomez force-pushed the debug-impl branch from 6f120d0 to 3283e71 Compare November 10, 2017 18:07

EPashkin reviewed Nov 10, 2017

View reviewed changes

Generate Debug for sys types

5a15289

GuillaumeGomez force-pushed the debug-impl branch from 3283e71 to 5a15289 Compare November 11, 2017 11:09

EPashkin reviewed Nov 11, 2017

View reviewed changes

Use debug_struct instead

54a102e

EPashkin reviewed Nov 11, 2017

View reviewed changes

sdroege reviewed Nov 11, 2017

View reviewed changes

GuillaumeGomez merged commit e58a8db into gtk-rs:master Nov 11, 2017

GuillaumeGomez deleted the debug-impl branch November 11, 2017 15:30

Uh oh!

Uh oh!

Conversation

GuillaumeGomez commented Nov 10, 2017

Uh oh!

GuillaumeGomez commented Nov 10, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

EPashkin commented Nov 10, 2017

Uh oh!

GuillaumeGomez commented Nov 10, 2017

Uh oh!

EPashkin commented Nov 10, 2017

Uh oh!

GuillaumeGomez commented Nov 11, 2017

Uh oh!

GuillaumeGomez commented Nov 11, 2017

Uh oh!

sdroege commented Nov 11, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

GuillaumeGomez commented Nov 11, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

EPashkin commented Nov 11, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sdroege commented Nov 11, 2017

Uh oh!

GuillaumeGomez commented Nov 11, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

GuillaumeGomez commented Nov 11, 2017

Uh oh!

GuillaumeGomez commented Nov 11, 2017

Uh oh!

EPashkin commented Nov 11, 2017

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

GuillaumeGomez commented Nov 11, 2017 •

edited

Loading