search

cbielow / PTXQC

A Quality Control (QC) pipeline for Proteomics (PTX) results generated by MaxQuant
Other
42 stars 25 forks source link

MissedCleavages plot missing #90

Closed cbielow closed 3 years ago

cbielow commented 3 years ago

in v1.0.7 the MC plot is not shown anymore, but to a bug introduced in Sep. 2020 here: https://github.com/cbielow/PTXQC/commit/803ebe3962d4e523dfddb54b2247ca851893d708

Thanks to Georg for reporting this!

cbielow commented 3 years ago

Solutions: wait for version 1.0.8 or revert to 1.0.4

fabianegli commented 3 years ago

Sorry to nag, but do you have a timeline for the release of version 1.0.8?

cbielow commented 3 years ago

I was planning to incorporate another thing I'm working on... bit this may only be ready in a month or so. You can also install via install_github(...) -- see installation page. This will give you the current development version, which contains the fix. If thats not an option, I'll do a release.

fabianegli commented 3 years ago

Thank you for the reply. Yes, I already already make use of the current master, just thought it would be nice to have this released.

On another note - maybe worth a separate issue - the version indicated form the current master is indicated in the report filename and it already uses the 1.0.8 version number although it obviously is something like 1.0.8dev.

Just to state the obvious. Anyone getting the Package from CRAN now will lack the missed cleavage report. Because it is an important part of the quality assessment I think it is worth releasing this simple bugfix fast instead of waiting for the implementation of new functionality, IMHO.

cbielow commented 3 years ago

On another note - maybe worth a separate issue - the version indicated form the current master is indicated in the report filename and it already uses the 1.0.8 version number although it obviously is something like 1.0.8dev.

Exactly my thoughts! However, the R package building code thinks that 1.0.8xxx is not a valid version number. It NEEDS semantic versioning (I recently forgot, and tried https://github.com/cbielow/PTXQC/commit/8fbef1f4f700df79af6d72d5a79789c7f40b08aam but then you cannot install the package). No more, no less. It's a shame.

Just to state the obvious. Anyone getting the Package from CRAN now will lack the missed cleavage report. Because it is an important part of the quality assessment I think it is worth releasing this simple bugfix fast instead of waiting for the implementation of new functionality, IMHO.

There is another side of the coin: CRAN maintainers will complain if you release too often (true story). So I tend to wait for at least 2 months, but I'll try a release soon.

fabianegli commented 3 years ago

Hmmm ... strange thing this R-verse. Thank you for your very informative reply and work on this package.

fabianegli commented 3 years ago

Well, after reading through some of the R package development documentation it becomes clear that they are not encouraging short release cycles because it means a lot of manual curation for them - which probably is intended to keep a high standard fro packages on CRAN. A two edged sword as seen in this case.

A thought on the release numbering. To prevent the "missing" dev versions, you could use even numbers on the patch level for releases that go to CRAN and uneven numbers for code that is in development or vice versa. Just a thought.

cbielow commented 3 years ago

Release 1.0.8 is on CRAN now (as well as on GitHub, obviously)

fabianegli commented 3 years ago

Thank you!

cbielow commented 3 years ago

A thought on the release numbering. To prevent the "missing" dev versions, you could use even numbers on the patch level for releases that go to CRAN and uneven numbers for code that is in development or vice versa. Just a thought.

with missing dev versions you mean versions missing from CRAN? Not sure that even/odd scheme would help there, because people would have to know about this to benefit from the approach. Ok, you can argue that the schema is kind of obvious... :) But what if I do two Github-only releases in a row? This would mean trouble... So for now, I'll just keep tagging each version as appropriate until we find a solution which has less drawbacks. Not a big problem at the moment IMHO.

fabianegli commented 3 years ago

With the missing dev versions I meant that you can't use versions like 1.0.0dev on CRAN. For two consecutive GitHub only releases you could just skip the in-between even number that would have been used for a release on CRAN.

The benefit of it would be that if there is a version number of a version released on CRAN in the report file name it can not be from a pre-release development version that carried the same version string.

fabianegli commented 3 years ago

Presumably, the vast majority of users is not affected by this versioning issue because they use the CRAN releases exclusively. This means it is not a very impactful issue, but on the other hand I think you only have to increase the version number by one before each release to CRAN which is simple and quick to do IMHO.