Enhancement: consider a sanity check on workdir and topdir before doing make clobber

[lcn2]

Is there an existing issue for this?

• I have searched for existing issues and did not find anything like this

Describe the enhancement

As mentioned by @TheEndCrystal28 in Discord:

There's an annoying feature where if you accidentally swap workdir and topdir in the arguments, the tool runs or in case both workdir/remarks.md and topdir/remarks.md(and fails) before checking that workdir and topdir are in the correct order.

Relevant images, screenshots or other files

No response

Relevant links

https://discord.com/channels/1327029557718417438/1332205820925972541/1476210418065084577

Anything else?

UPDATE 1e

We don't need to go overboard with this sanity check.

Assuming that topdir and workdir are different directories (see below):

If you find topdir/prog.c missing, AND you find workdir/prog.c exists, then assume the user has mixed up the args on the command line.

If you find topdir/remarks.md missing, AND you find workdir/remarks.md exists, then assume the user has mixed up the args on the command line.

Treat these cases as a "invalid command line". Issue an error message to stderr, print the usage message, and exit 3.

If no topdir/prog.c AND no workdir/prog.c are found, don't trigger an "invalid command line" error. Let the content checking code later on deal with that as it does now.

If no topdir/remarks.md AND no workdir/remarks.md are found, don't trigger an "invalid command line" error. Let the content checking code later on deal with that as it does now.

UPDATE 2b

As suggested below, this check will need to determine if topdir and workdir are different directories (i.e., use stat(2) and check if the dev_t and ino_t values are different).

If topdir and workdir are the same directories, then treat this as a "invalid command line". Issue an error message to stderr, print the usage message, and exit 3.

This check should be done before the prog.c and remarks.md checks.

UPDATE 3b

This check could be extended apply to checking Makefile as well.

An "invalid command line" error would also arise if you found workdir/Makefile and NOT topdir/Makefile.

It is OK if Makefile is found in BOTH topdir and workdir (after determining that topdir and workdir are different directories). The submitter might have a reason to use a workdir/Makefile.

If no topdir/Makefile AND if no workdir/Makefile are found, don't trigger an "invalid command line" error. Let the content checking code later on deal with that as it does now.

[xexyl]

Oh dear. This is not necessarily a bad idea.

I did check (not sure if it's still there) that the answers file is not put in the tarball but this might be worth considering too.

I am afraid I am much too tired to even consider ideas but you're welcome to assign this to me. Obviously for after this IOCCC. And who knows if I'll manage it before my surgery. Maybe during judging but then you might be absent from discussion at the time.

[xexyl]

First point: yes before any prompt in fact. Right after getting the absolute paths.

Presumably we'd use fts_walk().

The question is do we actually swap them? I personally think that would be a mistake. Again this is MY opinion but I believe that if we did that it would be making blatant assumptions and possibly destructive assumptions. Frankly I could see how running make clobber at all is destructive. It's probably pretty rare but I could imagine a scenario where someone accidentally ends up deleting their prog.c or other important files. That would be terrible.

As for the latter thing we could ask them to confirm before make clobber is executed IF that is an idea.

As for the former what MIGHT be better is to PROMPT the user if there APPEARS to be a mismatch. They can confirm and if they say no then exit with an error. BUT there is an equal argument for making it an error anyway since it IS a command line error. On the other hand ... if they don't mean it that way it could be annoying.

What I do not understand however is why this is even an issue now I think on it. The code actually checks if prog.c is there and it wouldn't be in the workdir so how is this even possible (maybe they meant something else though?)?

Before working on this I would guess these questions and thoughts should be considered.

[lcn2]

First point: yes before any prompt in fact. Right after getting the absolute paths.

Presumably we'd use fts_walk().

The question is do we actually swap them? I personally think that would be a mistake. Again this is MY opinion but I believe that if we did that it would be making blatant assumptions and possibly destructive assumptions. Frankly I could see how running make clobber at all is destructive. It's probably pretty rare but I could imagine a scenario where someone accidentally ends up deleting their prog.c or other important files. That would be terrible.

Please do NOT swap them.

As for the latter thing we could ask them to confirm before make clobber is executed IF that is an idea.

As for the former what MIGHT be better is to PROMPT the user if there APPEARS to be a mismatch. They can confirm and if they say no then exit with an error. BUT there is an equal argument for making it an error anyway since it IS a command line error. On the other hand ... if they don't mean it that way it could be annoying.

Perform a simple "sanity check" on the two directories, where the key is simple. The check does NOT need to be exhaustive: there isn't a need to check for every possible indicator that the user has the args in the wrong order.

Look for a few of the required workdir files such as:

• Makefile
• prog.c
• remarks.md

If you find any of those sitting directly under topdir, AND you find them missing from workdir, then assume the user mixed up the arg order, AND issue an error message, AND exit with a non-zero exit code. Such a condition would fail anyway due to the lack of required files in a submission, so bailing out with an error is fine.

BTW: The error message might suggest that the workdir topdir args are in the wrong order.

Please consider submitting a PR, @xexyl, to resolve this issue, so that the PR may be considered after the winners of IOCCC29 are published.

[lcn2]

Schedule notice

About two or three weeks after the IOCCC29 winning entries
have been posted, and we process some of the early
pull requests
against the 2025 directory tree, the
IOCCC Judges plan to go on an IOCCC vacation.

We had intended to go on an IOCCC vacation after releasing the winners of
IOCCC28, but then the efforts to process bug fixes and enhancements
to the mkiocccentry repo
took so much time that by the time that repo was stable, it was time to
open IOCCC29. Therefore, this time, we plan to wait until after
the end of our post-IOCCC29 IOCCC vacation before working on any
mkiocccentry repo PRs.

Folks will have time to refine any PRs and/or to submit new PRs against the
mkiocccentry repo. We ask for your patience in advance.

[xexyl]

First point: yes before any prompt in fact. Right after getting the absolute paths.
Presumably we'd use fts_walk().
The question is do we actually swap them? I personally think that would be a mistake. Again this is MY opinion but I believe that if we did that it would be making blatant assumptions and possibly destructive assumptions. Frankly I could see how running make clobber at all is destructive. It's probably pretty rare but I could imagine a scenario where someone accidentally ends up deleting their prog.c or other important files. That would be terrible.

Please do NOT swap them.

You misunderstood me.

I meant: if the user input the wrong order do we issue a fatal error, warning or do we actually swap the strings so that they can proceed?

I can see how the latter is too presumptuous. But on the other hand the former might be annoying if they have some strange requirement?

As for the latter thing we could ask them to confirm before make clobber is executed IF that is an idea.
As for the former what MIGHT be better is to PROMPT the user if there APPEARS to be a mismatch. They can confirm and if they say no then exit with an error. BUT there is an equal argument for making it an error anyway since it IS a command line error. On the other hand ... if they don't mean it that way it could be annoying.

Perform a simple "sanity check" on the two directories, where the key is simple. The check does NOT need to be exhaustive: there isn't a need to check for every possible indicator that the user has the args in the wrong order.

Look for a few of the required workdir files such as:

• Makefile
• prog.c
• remarks.md

If you find any of those sitting directly under topdir, AND you find them missing from workdir, then assume the user mixed up the arg order, AND issue an error message, AND exit with a non-zero exit code. Such a condition would fail anyway due to the lack of required files in a submission, so bailing out with an error is fine.

Hmm ... but then it should (as suggested above and below) ask them rather than be an error. We should not assume these details surely?

As for the code to do this it'll take more time and thought and more than all that energy, for me to do this.

I however have fixed a number of the current issues and will soon (I hope) open a single pull request (since there is no actual code change except that various tools now print the repo version as well).

BTW: The error message might suggest that the workdir topdir args are in the wrong order.

Okay so you think error message. But...shouldn't it ask them first?

[lcn2]

First point: yes before any prompt in fact. Right after getting the absolute paths.
Presumably we'd use fts_walk().
The question is do we actually swap them? I personally think that would be a mistake. Again this is MY opinion but I believe that if we did that it would be making blatant assumptions and possibly destructive assumptions. Frankly I could see how running make clobber at all is destructive. It's probably pretty rare but I could imagine a scenario where someone accidentally ends up deleting their prog.c or other important files. That would be terrible.

Please do NOT swap them.

You misunderstood me.

Sorry (tm Canada 🇨🇦)

I meant: if the user input the wrong order do we issue a fatal error, warning or do we actually swap the strings so that they can proceed?

How about a fatal error and exit non-zero?

[lcn2]

BTW: The error message might suggest that the workdir topdir args are in the wrong order.

Okay so you think error message. But...shouldn't it ask them first?

Perhaps the error message could state by running with --arg-order-is-correct (or something like that), the user could force the order and by pass this error message (turning it into a warning and not exiting)?

[xexyl]

First point: yes before any prompt in fact. Right after getting the absolute paths.
Presumably we'd use fts_walk().
The question is do we actually swap them? I personally think that would be a mistake. Again this is MY opinion but I believe that if we did that it would be making blatant assumptions and possibly destructive assumptions. Frankly I could see how running make clobber at all is destructive. It's probably pretty rare but I could imagine a scenario where someone accidentally ends up deleting their prog.c or other important files. That would be terrible.

Please do NOT swap them.

You misunderstood me.

Sorry (tm Canada 🇨🇦)

Not a problem. I could have been clearer too.

I meant: if the user input the wrong order do we issue a fatal error, warning or do we actually swap the strings so that they can proceed?

How about a fatal error and exit non-zero?

I'll get to this in your later comment.