Disallow overwrite_key and force_update in Config File #218

anipaul2 · 2023-05-11T19:39:58Z

fixes #217

anipaul2 · 2023-05-11T20:10:28Z

I have added the complete line and pushed it but only half gets pushed. What maybe the possible error?

mariocynicys · 2023-05-11T20:12:43Z

save the file maybe 🤔

anipaul2 · 2023-05-11T20:16:12Z

save the file maybe 🤔

Done. Actually, I thought it does it automatically but saw now I have not enabled the feature. Showing now.

sr-gi · 2023-05-12T04:28:21Z

.gitignore

+.idea
+.DS_Store


Don't add this change to the PR, these are entirely unrelated.

When I raise a pr, it gets attached automatically and pr gets two commits. So, rebased it and squashed it. What Should I do to make the commit ignore in this pr?

Delete it, amend and rebase so it is a single commit

Delete it, amend and rebase so it is a single commit

okay.

sr-gi · 2023-05-12T04:33:56Z

teos/src/config.rs

@@ -203,6 +203,12 @@ impl Config {
        self.tor_support |= options.tor_support;
        self.debug |= options.debug;
        self.deps_debug |= options.deps_debug;
+        if self.overwrite_key {
+            panic!("The 'overwrite_key' parameter cannot be set in the configuration file. Please pass it as a command-line argument if necessary.");


Don't panic. Make the method return a Result and check it on main, if it's an error, eprintln!("{e}") and std::process::exit(1)

Don't panic. Make the method return a Result and check it on main, if it's an error, eprintln!("{e}") and std::process::exit(1)

if path_with_options have to updated with result, I have to add Result<(), String> in the function to make the error message work. But the problem is I have to change all the code path function to return a result.

Don't panic. Make the method return a Result and check it on main, if it's an error, eprintln!("{e}") and std::process::exit(1)

You should do such sanity checks in config::from_file instead of patch_with_options.

You should do such sanity checks in config::from_file instead of patch_with_options.

config::from_file is in main.rs where I don't see any changes can be made regarding this issue. In config.rs, from_file function can be modified to return a Result<T, String> instead of T.

anipaul2 · 2023-05-12T20:57:14Z

I made changes only in the main.rs file because modifying the config.rs file's from_file function would have resulted in errors. The reason for this is that when we removed the generic type T and added Config directly, it caused errors related to the opt variable being linked with overwrite_key and force_update.

Since opt is specific to the command line options, it cannot be directly used in the from_file function, which is responsible for loading the configuration from a file. Therefore, I decided to handle the sanity checks in the main.rs file instead.

mariocynicys

You should have these checks inside config.rs.
What you could do is modify the from_filefunction like so:

pub fn from_file<T: Default + serde::de::DeserializeOwned>(path: PathBuf) -> Result<T, String> {
    match std::fs::read(path) {
        Ok(file_content) => match toml::from_slice::<T>(&file_content) {
            Err(e) => {
                Err(format!("Couldn't parse config file: {e}"))
            }
            Ok(config) => {
                // do the sanity checks here.
                if config.overwrite_key {
                    return Err(format!("Overwrite key isn't allowed in the config file."))
                }
                Ok(config)
            }
        },
        Err(_) => Ok(T::default()),
    }
}

This won't work right away, you will need to adapt some stuff in main.rs and probably cli.rs because it uses the same parsing function.

mariocynicys · 2023-05-16T09:37:23Z

teos/src/main.rs

+    if conf.overwrite_key {
+        eprintln!("The 'overwrite_key' parameter cannot be set in the configuration file. Please pass it as a command-line argument if necessary.");
+        std::process::exit(1);
+    }
+    if conf.force_update {
+        eprintln!("The 'force_update' parameter cannot be set in the configuration file. Please pass it as a command-line argument if necessary.");
+        std::process::exit(1);
+    }


This doesn't belong here.

anipaul2 · 2023-05-18T18:16:20Z

You should have these checks inside config.rs. What you could do is modify the from_filefunction like so:

pub fn from_file<T: Default + serde::de::DeserializeOwned>(path: PathBuf) -> Result<T, String> {
    match std::fs::read(path) {
        Ok(file_content) => match toml::from_slice::<T>(&file_content) {
            Err(e) => {
                Err(format!("Couldn't parse config file: {e}"))
            }
            Ok(config) => {
                // do the sanity checks here.
                if config.overwrite_key {
                    return Err(format!("Overwrite key isn't allowed in the config file."))
                }
                Ok(config)
            }
        },
        Err(_) => Ok(T::default()),
    }
}

This won't work right away, you will need to adapt some stuff in main.rs and probably cli.rs because it uses the same parsing function.

Got it.

anipaul2 · 2023-05-19T17:45:45Z

You should have these checks inside config.rs. What you could do is modify the from_filefunction like so:

pub fn from_file<T: Default + serde::de::DeserializeOwned>(path: PathBuf) -> Result<T, String> {
    match std::fs::read(path) {
        Ok(file_content) => match toml::from_slice::<T>(&file_content) {
            Err(e) => {
                Err(format!("Couldn't parse config file: {e}"))
            }
            Ok(config) => {
                // do the sanity checks here.
                if config.overwrite_key {
                    return Err(format!("Overwrite key isn't allowed in the config file."))
                }
                Ok(config)
            }
        },
        Err(_) => Ok(T::default()),
    }
}

This won't work right away, you will need to adapt some stuff in main.rs and probably cli.rs because it uses the same parsing function.

I've tried implementing the changes you've suggested. Specifically, I updated the from_file function to include checks for overwrite_key. However, I'm encountering an issue. The from_file function is a generic function, which takes any type T that implements Default + serde::de::DeserializeOwned. This means it can't access fields specific to Config like overwrite_key directly, because these fields aren't part of the constraints we've put on T. This seems to lead to a type mismatch. I tried updating main.rs and cli.rs to adapt to these changes as well, but the issue persists. Could you kindly provide some guidance on how I could implement the desired checks while preserving the generic nature of the from_file function?

I am thinking of removing the T and make it config.

mariocynicys · 2023-05-22T15:10:20Z

I think a good solution is to serialize the config as a serde object and disallowing the fields we want disallowed. @anipaul2

anipaul2 · 2023-05-23T16:28:35Z

I think a good solution is to serialize the config as a serde object and disallowing the fields we want disallowed. @anipaul2

Yes sounds like a reasonable approach.

…d force_update fields from serialization. By using the #[serde(skip_serializing)] attribute, these fields will not be included when the Config struct is serialized.

anipaul2 · 2023-05-23T18:17:12Z

@sr-gi @mariocynicys I have first made the changes in cli.rs, main.rs, and config.rs to disallow overwrite_key and force_update but only skipping the serializing in the flags part does this. Let me know if this approach is wrong but it worked on my end.

sr-gi · 2023-05-23T21:01:09Z

I don't think what you are doing is right. This is only preventing Coinf from serializing those two items. You could prevent it from deserializing it, which will prevent us from reading that, but I don't think that'll be correct either.

Giving this a closer look, I'm starting to lean toward two different solutions:

Either reworking the Config logic to specify where the data comes from, potentially using an enum, which will help towards #204 (comment), or doing this either with a custom deserializer or in patch_with_options.

I think @anipaul2 has a point in that from_file may not be the best place to handle this, given that is used for both reading from teosd conf and teos-cli conf, and these two params have nothing to do with the latter.

mariocynicys · 2023-05-24T10:15:09Z

given that is used for both reading from teosd conf and teos-cli conf, and these two params have nothing to do with the latter.

Agreed. Even though these two parameter should just default to None when queried in the cli config. But doesn't make a lot of sense querying it in the first place.

Either reworking the Config logic to specify where the data comes from, potentially using an enum, which will help towards #204 (comment), or doing this either with a custom deserializer or in patch_with_options.

Not sure how the enum would help regarding the comment you linked. I'm thinking now of copying this function over to cli_config and not make it generic anymore. Could this help the issue you linked?

sr-gi · 2023-05-24T19:39:25Z

given that is used for both reading from teosd conf and teos-cli conf, and these two params have nothing to do with the latter.

Agreed. Even though these two parameter should just default to None when queried in the cli config. But doesn't make a lot of sense querying it in the first place.

Either reworking the Config logic to specify where the data comes from, potentially using an enum, which will help towards #204 (comment), or doing this either with a custom deserializer or in patch_with_options.

Not sure how the enum would help regarding the comment you linked. I'm thinking now of copying this function over to cli_config and not make it generic anymore. Could this help the issue you linked?

This is what I'm thinking about give or take. However, I haven't been able to make the deserialization of ConfigParam work, looks like there's an issue with the trait bounds that I'm still scratching my head about:

master...sr-gi:rust-teos:config-rework

mariocynicys · 2023-05-24T19:51:30Z

looks like there's an issue with the trait bounds that I'm still scratching my head about

@sr-gi Fixed the compilation issues in this branch. Acking the approach :).
https://github.com/mariocynicys/rust-teos/tree/config-rework-fix

anipaul2 · 2023-06-04T08:12:06Z

@sr-gi @mariocynicys how to approach this issue now? any ideas? I am thinking of making a change in patch_with_options and defining a new struct for overwrite_key and force_update.

sr-gi · 2023-06-05T14:04:19Z

I think I'd make more sense to fix this one by reworking the config as I was trying to do here: master...sr-gi:rust-teos:config-rework

I haven't had much time lately to finish it, so no worries about this for now, it's not critical, or you can pick up when I left it if you like

anipaul2 · 2023-06-05T14:46:04Z

I think I'd make more sense to fix this one by reworking the config as I was trying to do here: master...sr-gi:rust-teos:config-rework

I haven't had much time lately to finish it, so no worries about this for now, it's not critical, or you can pick up when I left it if you like

okay.

anipaul2 force-pushed the issue217 branch from e776708 to 602ec86 Compare May 11, 2023 19:54

anipaul2 force-pushed the issue217 branch from 602ec86 to a27e1a6 Compare May 11, 2023 20:14

sr-gi requested changes May 12, 2023

View reviewed changes

anipaul2 force-pushed the issue217 branch from a27e1a6 to 5e97fce Compare May 12, 2023 20:53

anipaul2 requested review from sr-gi and mariocynicys May 15, 2023 07:00

mariocynicys requested changes May 16, 2023

View reviewed changes

anipaul2 requested a review from mariocynicys May 21, 2023 13:47

anipaul2 force-pushed the issue217 branch from 5e97fce to ebb86b6 Compare May 23, 2023 18:10

This commit updates the Config struct to exclude the overwrite_key an…

61567b7

…d force_update fields from serialization. By using the #[serde(skip_serializing)] attribute, these fields will not be included when the Config struct is serialized.

anipaul2 force-pushed the issue217 branch from ebb86b6 to 61567b7 Compare May 23, 2023 18:14

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Disallow overwrite_key and force_update in Config File #218

Disallow overwrite_key and force_update in Config File #218

anipaul2 commented May 11, 2023 •

edited

anipaul2 commented May 11, 2023

mariocynicys commented May 11, 2023

anipaul2 commented May 11, 2023

sr-gi May 12, 2023

anipaul2 May 12, 2023

sr-gi May 12, 2023

anipaul2 May 12, 2023

sr-gi May 12, 2023

anipaul2 May 12, 2023

anipaul2 May 12, 2023

mariocynicys May 12, 2023

anipaul2 May 12, 2023

anipaul2 commented May 12, 2023 •

edited

mariocynicys left a comment •

edited

mariocynicys May 16, 2023

anipaul2 commented May 18, 2023

anipaul2 commented May 19, 2023 •

edited

mariocynicys commented May 22, 2023

anipaul2 commented May 23, 2023

anipaul2 commented May 23, 2023

sr-gi commented May 23, 2023

mariocynicys commented May 24, 2023

sr-gi commented May 24, 2023

mariocynicys commented May 24, 2023

anipaul2 commented Jun 4, 2023

sr-gi commented Jun 5, 2023

anipaul2 commented Jun 5, 2023

		.idea
		.DS_Store

Disallow overwrite_key and force_update in Config File #218

Are you sure you want to change the base?

Disallow overwrite_key and force_update in Config File #218

Conversation

anipaul2 commented May 11, 2023 • edited

anipaul2 commented May 11, 2023

mariocynicys commented May 11, 2023

anipaul2 commented May 11, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

anipaul2 commented May 12, 2023 • edited

mariocynicys left a comment • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

anipaul2 commented May 18, 2023

anipaul2 commented May 19, 2023 • edited

mariocynicys commented May 22, 2023

anipaul2 commented May 23, 2023

anipaul2 commented May 23, 2023

sr-gi commented May 23, 2023

mariocynicys commented May 24, 2023

sr-gi commented May 24, 2023

mariocynicys commented May 24, 2023

anipaul2 commented Jun 4, 2023

sr-gi commented Jun 5, 2023

anipaul2 commented Jun 5, 2023

anipaul2 commented May 11, 2023 •

edited

anipaul2 commented May 12, 2023 •

edited

mariocynicys left a comment •

edited

anipaul2 commented May 19, 2023 •

edited