MOBILE_H syntax error observed with v1.07 I/O validation

[baleland]

CMD> inchi-1.exe ptable.sdf /FixedH /RecMet
CMD> inchi-1.exe ptable.sdf.txt /InChI2InChI /FixedH /RecMet

results in,

Structure: 1 Syntax error (-2) in MOBILE_H, Reconnected layer (102)

Structure: 2 Syntax error (-2) in MOBILE_H, Reconnected layer (102)

Structure: 3 Syntax error (-2) in MOBILE_H, Reconnected layer (102)

I am loathe to restandardize tests generally w/o a firm understanding. Here the inchi utility itself is emitting output that cannot be reparsed w/o error, which seems generally wrong.

This is a test standard we have run successfully v1.05 & v1.06, which now fails.

ptable.sdf.txt

[djb-rwth]

Hi @baleland,
I can confirm this output and will mark it as a bug.
Will get back with the fix shortly.

[djb-rwth]

Hi @baleland,
I think I am very near fixing this issue.
Please let me know what is your deadline and if you think that this fix might resolve the testing problems you are facing.

[baleland]

we are in the beta cycle, so probably ~1mo away from a release candidate lockdown. while this fix (& the others) would be required, there is still the issue(s) of a number of tests that differ from v1.06 - unclear that we would move to v1.07.1/2 unless those differences are fully understood.

[djb-rwth]

Hi @baleland,
Thanks for the information.
I think I might have fixed this issue as well.
Please swap the ichiread.c with the one attached to this message and do let me know if this issue has been fixed.
The problem was similar to the one we experienced in issue #27 and #28.
ichiread.c.txt

[baleland]

much more progress made now!! All the InChI tests were now able to be run.
Only ONE test "fails" due to differences b/w the v1.06 and v1.07.1.x results - maybe due to needing restandardization, or maybe due to and actual/unintended difference,

***KNOWN FAILURES: 0
*** NEW  FAILURES: 262

would you like me to hand select some representative examples from these 262 failures and open a new case for analysis/verifcation?

[djb-rwth]

much more progress made now!! All the InChI tests were now able to be run. Only ONE test "fails" due to differences b/w the v1.06 and v1.07.1.x results - maybe due to needing restandardization, or maybe due to and actual/unintended difference,

Hi @baleland,
Thanks for the great news.

***KNOWN FAILURES: 0
*** NEW  FAILURES: 262

would you like me to hand select some representative examples from these 262 failures and open a new case for analysis/verifcation?

Yes, absolutely. I will close this issue and you can open a new one now.

[baleland]

InChI_1_07_0\bugfix\inchi-1.exe STATUS_7_IO_ERROR.txt /InChI2InChI /FixedH /RecMet

! Working in engineering mode


Structure: 1 Syntax error (-2) in MOBILE_H (2)
Finished processing 1 structure: 0 errors, processing time 0:00:00.00

we see 25 I/O failures of this type in our tests (with your 2x module updates) - the bugfix exe you provided previously does not have your latest fixes.

STATUS_7_IO_ERROR.txt

[baleland]

similarly for this set of structures, we see 237 I/O failures. I do not have an inchi-1.exe with your updates to compare with, but there are at least 59 errors from the last version you provided.

STATUS_8_EXAMPLES.txt

[djb-rwth]

Hi @baleland,
I reopened this issue so we can take it from here.
Will get back to you shortly.

[djb-rwth]

Hi @baleland,
It looks as if all these tests have passed with the fixes I applied above.
I am sending you an .exe version for Windows and elf file for Ubuntu/Linux in the attachment along with the test results.
Please let me know if the tests pass with these executables at your end.
If they do, I will upload a new version to our rwth branch, i.e. both the source codes and the executables.
v_1.07.2_rc.zip

[baleland]

Ok, I guess I'll have to walk through some of the test failures to see why the test results are different than the inchi-1.exe I/O.

[baleland]

Ok - I've done a thorough review of our existing tests. One particular test (rarely run) "appeared" to show a difference b/w v1.06 and v1.07 I/O round-trip testing via /FixedH /RecMet.

In reality, it appears this test had been disabled and the v1.06 output has the same number of I/O errors as v1.07. The errors themselves are likely due to our testing code, nothing to do with the InChI engine itself, seemingly.

[djb-rwth]

Hi @baleland,
Thank you for the report.
Please do keep me updated on this.

[flange-ipb]

Thomas Exner submitted a very similar bug report that is resolved by the patched executables provided by @djb-rwth in #43 (comment).

inchi-1 input.inchi -InChI2Struct -OutputSDF fails with

Structure: 1 Syntax error (-2) in MOBILE_H (2)

for the input InChI InChI=1S/C6H8O7.3Na.2H2O/c7-3(8)1-6(13,5(11)12)2-4(9)10;;;;;/h13H,1-2H2,(H,7,8)(H,9,10)(H,11,12);;;;2*1H2/q;3*+1;;/p-3 (Sodium citrate (dihydrate)).

[djb-rwth]

Hi @flange-ipb,
Thank you for confirming this.
In that case, it seems that we have fixed all major bugs which will be added in v1.07.2.

Still waiting for @baleland's confirmation that there are no new issues at his end.

[baleland]

I have no new failures to report - awaiting the 1.07.2 official release.

…

________________________________ From: Djordje Baljozovic ***@***.***> Sent: Saturday, September 21, 2024 5:56 AM To: IUPAC-InChI/InChI ***@***.***> Cc: Burt Leland ***@***.***>; Mention ***@***.***> Subject: Re: [IUPAC-InChI/InChI] MOBILE_H syntax error observed with v1.07 I/O validation (Issue #43) Hi @flange-ipb<https://github.com/flange-ipb>, Thank you for confirming this. In that case, it seems that we have fixed all major bugs which will be added in v1.07.2. Still waiting for @baleland<https://github.com/baleland>'s confirmation that there are no new issues at his end. — Reply to this email directly, view it on GitHub<#43 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ACDX2GUEWKRBWZZNMPCUT23ZXVNITAVCNFSM6AAAAABMMQKIKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNRVGE2TSNZSGQ>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[djb-rwth]

Hi @baleland,
All fixes to issues we discussed in your posts have been integrated in InChI v1.07.2, which has now been uploaded to rwth branch.
Please feel free to let me know if you have any further suggestions or comments in this regard.
Hoping to hear from you soon.

[baleland]

Ok – thanks for the info. Let me know when this would become a release branch – we cannot use anything but release branches for our Tk releases. From: Djordje Baljozovic ***@***.***> Sent: Monday, November 11, 2024 4:03 PM To: IUPAC-InChI/InChI ***@***.***> Cc: Burt Leland ***@***.***>; Mention ***@***.***> Subject: Re: [IUPAC-InChI/InChI] MOBILE_H syntax error observed with v1.07 I/O validation (Issue #43) Hi @baleland<https://github.com/baleland>, All fixes to issues we discussed in your posts have been integrated in InChI v1.07.2, which has now been uploaded to rwth<https://github.com/IUPAC-InChI/InChI/tree/rwth> branch. Please feel free to let me know if you have any further suggestions or comments in this regard. Hoping to hear from you soon. — Reply to this email directly, view it on GitHub<#43 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ACDX2GX2OB7HJM4BMFDLGGD2AEZRFAVCNFSM6AAAAABMMQKIKKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDINRZGI2TGMJSGQ>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[djb-rwth]

Hi @baleland,
No problems -- will let you know as soon as v.1.07.2 release candidate becomes an official release, i.e. is uploaded to main branch.