The `[patch]` approach causes the tree not to build when used as a
dependency, unless the `[patch]` is replicated into the depending
project.
Instead, replace our `derive_builer =` dependencies with a reference
to a specific git commit:
perl -i~ -pe 'next unless m/^derive_builder/; s#"(0\.11\.2)"#{ version = "$1", git = "https://github.com/ijackson/rust-derive-builder", rev = "ba0c1a5311bd9f93ddf5f5b8ec2a5f6f03b22fbe" }#' crates/*/Cargo.toml
Note that the commitid has changed. This is because derive_builder is
in fact a workspace of 4 crates. 3 of them are of interest to arti
itself (the 4th exists only for testing). So the same "add git
revision" treatment had to be done to the `derive_builder` and
`derive_builder_macro` crates. Each dependency edge involves a new
commit in the derive_builder workspace, since we can't create a git
commit containing its own commitid. (We want to use commits, rather
than a branch, so that what we are depending on is actually properly
defined, and not subject to the whims of my personal github
namespace.)
There are no actual code changes in derive_builder.
The arti crate itself is only used in benchmark and testing crate.
I think this sentence does not belong here.
Also extend retry-error description (from Architecture.md).
Now that we require Rust 1.56, we can upgrade to AES 0.8. This
forces us to have some slight API changes.
We require cipher 0.4.1, not cipher 0.4.0, since 0.4.0 has
compatibility issues with Rust 1.56.
This is actually a number of *attempts* not a number of *retries*.
The setter method was already called "attempts".
This chnages the deserialisation of the config.
Use sub_builder. We must do something special for defaults.
This involves moving the actual default values for retry_bootstrap and
retry_microdescs into config.rs, since they need to access the fields
of the un-built version of the structure. (An alternative would be to
generate "weak setters" which do not override previous settings, but
derive_builder does not offer to generate them and that seems
overkill.)
Instead, everyone should use DownloadScheduleBuilder.
The new() method would in any case be useless in a moment, since we're
going to embed DownloadScheduleBuilder in the NetworkConfig, not
DownloadSchedule.
The call sites in the tests are all about to change again.
The current behaviour is to treat 0 as indicating "use the default",
which is quite strange. We are going to get rid of that.
The new way will be to reject zero, during
DownloadScheduleBuilder::build, Add a test case for that.
Really, we probably don't want any of these not to be pub, but it
triggers "unreachable pub" in my test cases, and making it not pub by
mistake seems not very serious, and likely to be noticed.
Making the struct private in the test cases has the useful effect of
checking that all the methods are tested.
And add an imprecation in define_list_config_builder's doc comment do
do so in future for other invocations of the macro.
Add add the missing full stops.
This is an automated change made with a perl one-liner and verified
with grep -L and grep -l.
Some warnings are introduced with this change; they will be removed
in subsequent commits.
See arti#208 for older discussion on this issue.
Our support policy says that we can update to any Rust released at
least 6 months ago; 1.56 came out on 21 October 2021.
This doesn't yet change any code: it just increases the version
we say we need in our README, and the version we test against in
CI.
Our main justification for this change is to be able to upgrade to
newer versions of our dependencies, including `async_executors` >=
0.5, `aes` >= 0.8, and `cipher` >= 0.4.
This means that `NetworkConfig::initial_predicted_ports` is now like
the other list-like things, returning `&mut list_builder` with the same
`set()` and `append()` methods.
eta
Ian Jackson
Samanta Navarro
Nick Mathewson
Ian Jackson