make builds deterministic (#392)

This was just for testing. 
I think we might want some more control.

The `append_file` function wasn't actually using the prepared header so
the `mtime` was still set from the file system. Changing that to the
`tar::HeaderMode::Deterministic` works. I think we might still want more
control. The Deterministic header from the tar library does something to
the `chmod` bits and also sets the mtime to the same constant that I
copied out because that is the birthdate of Rust and some programs don't
handle a mtime of `0` not well apparently.

crates/rattler_conda_types/src/package/archive_type.rs

@@ -1,7 +1,10 @@
 use std::path::Path;
 
+use serde::{Deserialize, Serialize};
+
 /// Describes the type of package archive. This can be derived from the file extension of a package.
-#[derive(Copy, Clone, Debug, Ord, PartialOrd, Eq, PartialEq, Hash)]
+#[derive(Copy, Clone, Debug, Ord, PartialOrd, Eq, PartialEq, Hash, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
 pub enum ArchiveType {
     /// A file with the `.tar.bz2` extension.
     TarBz2,

crates/rattler_package_streaming/src/write.rs

@@ -229,22 +229,15 @@ fn prepare_header(
     header.as_gnu_mut().unwrap().name[..name.len()].clone_from_slice(&name[..]);
 
     let stat = fs::symlink_metadata(path)?;
-    header.set_metadata(&stat);
-
-    // erase some fields
-    header.set_uid(0);
-    header.set_gid(0);
-    header.set_device_minor(0)?;
-    header.set_device_major(0)?;
+    header.set_metadata_in_mode(&stat, tar::HeaderMode::Deterministic);
 
     if let Some(timestamp) = timestamp {
         header.set_mtime(timestamp.timestamp().unsigned_abs());
+    } else {
+        // 1-1-2023 00:00:00 (Fixed date in the past for reproducible builds)
+        header.set_mtime(1672531200);
     }
 
-    // let file_size = stat.len();
-    // TODO do we need this
-    //  + 1 to be compliant with GNU tar
-    // header.set_size(file_size + 1);
     Ok(header)
 }
 
@@ -264,10 +257,10 @@ fn append_path_to_archive(
         .map_err(|err| trace_file_error(&base_path.join(path), err))?;
 
     if header.entry_type().is_file() {
-        let mut file = fs::File::open(base_path.join(path))
+        let file = fs::File::open(base_path.join(path))
             .map_err(|err| trace_file_error(&base_path.join(path), err))?;
 
-        archive.append_file(path, &mut file)?;
+        archive.append_data(&mut header, path, &file)?;
     } else if header.entry_type().is_symlink() || header.entry_type().is_hard_link() {
         let target = fs::read_link(base_path.join(path))
             .map_err(|err| trace_file_error(&base_path.join(path), err))?;

crates/rattler_virtual_packages/src/lib.rs

@@ -157,7 +157,6 @@ fn try_detect_virtual_packages() -> Result<Vec<VirtualPackage>, DetectVirtualPac
 #[derive(Clone, Eq, PartialEq, Hash, Debug, Deserialize)]
 pub struct Linux {
     /// The version of linux
-    /// #[serde(deserialize_with = "from_str")]
     pub version: Version,
 }
 
@@ -194,7 +193,6 @@ pub struct LibC {
     pub family: String,
 
     /// The version of the libc distribution.
-    /// #[serde(deserialize_with = "from_str")]
     pub version: Version,
 }
 
@@ -230,7 +228,6 @@ impl From<LibC> for VirtualPackage {
 #[derive(Clone, Eq, PartialEq, Hash, Debug, Deserialize)]
 pub struct Cuda {
     /// The maximum supported Cuda version.
-    /// #[serde(deserialize_with = "from_str")]
     pub version: Version,
 }
 
@@ -316,7 +313,6 @@ impl From<Archspec> for VirtualPackage {
 #[derive(Clone, Eq, PartialEq, Hash, Debug, Deserialize)]
 pub struct Osx {
     /// The OSX version
-    /// #[serde(deserialize_with = "from_str")]
     pub version: Version,
 }