#259 io-2.4.2

[Philip Nienhuis]

Here is io-2.4.2.
Carnë, I've introduced a construct to get rid of the "octave_config_info deprecated" warning.
io is supported up to Octave-2.4.2 - I doubt if the spreadsheet I/O will make it into Octave-2.4.2.

md5sums:

5d2d959c05f4d90cf2400849f30354a9 io-2.4.2.tar.gz
6b0378a711f82b473c42d7efa2fdd273 io-html.tar.gz

Best wishes, Philip

[Carnë Draug]

You must have uploaded the wrong version of the package. The checksums do not match.

[Philip Nienhuis]

Well, I rather copied a wrong version of the md5sums file :-(

In the mean time I spotted release date errors in DESCRIPTION and NEWS => fixed.

new package version + html attached.

$ cat md5sums_io_2.4.2
ff92ba8273ab11bd4edd3a4a41b4808d io-2.4.2.tar.gz
7808db437422d68ab654b53c8d5f9de7 io-html.tar.gz

[Carnë Draug]

I'm confused. Why do you have dependencies octave < 4.4.0 ?

[Carnë Draug]

Also, to avoid copying the wrong version next time, you could write a Makefile like the ones in the other packages.

[Philip Nienhuis]

Q.: Why < 4.4.0?
A.: Same reason as 2 or 3 io pkg releases ago: there are still plans to move spreadsheet + XML I/O into core. I don't think it'll happen anymore or 4.2.0 but 4.4.0 makes a good chance. And if that happens I'd like to avoid shadowing of new core functions by the io pkg versions.
Based on past experience I expect another 2 or maybe 3 io pkg releases from now before 4.4.0 is there.

I'm not a fan of Makefiles if they aren't strictly needed for the package contents themselves. I copied the wrong md5sums, not the wrong package. I don't think a Makefile would have precluded that mistake.

[Carnë Draug]

Q.: Why < 4.4.0?
A.: Same reason as 2 or 3 io pkg releases ago: there are still plans to move spreadsheet + XML I/O into core. I don't think it'll happen anymore or 4.2.0 but 4.4.0 makes a good chance. And if that happens I'd like to avoid shadowing of new core functions by the io pkg versions.

Considering where we are now, that was obviously a mistake. Why make it again? And if core does implement those functions what are you going to do? Remove the xls functions from the package and force users of old Octave versions to fetch old versions of the package? When those functions move to core you should have a better strategy than blocking install by limitting versions on the dependencies (like what the mapping package is already doing for deg2rad and rad2deg).

I'm not a fan of Makefiles if they aren't strictly needed for the package contents themselves. I copied the wrong md5sums, not the wrong package. I don't think a Makefile would have precluded that mistake.

It would have avoided the mistake because the existing Makefiles for releases will clobber previous tarballs on the target directory and will print the md5sum.

[Philip Nienhuis]

Considering where we are now, that was obviously a mistake"
What mistake? Would you please explain what went wrong?

"If core does implement those functions what are you going to do?"
By that time we are 2 or 3 io package versions further. The then next io package won't have those functions and it will install only on the new core Octave. Older Octave will continue to work fine with the older io packages.

Remove the xls functions from the package and force users of old Octave versions to fetch old versions of the package?
Of course, what else? Where would the newest xls functions be then? maybe in core Octave? Would you want those new core functions copied into a new OF package version? Are there any examples?
If the spreadsheet functions still won't make it into core, no problem, they would be maintained in the newest io packages with a dependency on the then next major Octave version. If needed until JWE himself declares he doesn't want them in core.
Newest Octave requires newest OF packages. Nothing wrong with that.
Older Octave is just that - older Octave, with concomitant limitations. I sympathize with trying to offer the best and most recent functionality for older Octave as well, but not at all cost.

Although it is absolutely possible that I overlook something, I can't see it. Enlighten me please.

"It would have avoided the mistake because [....] will print the md5sum" -
It won't preclude accidentally scrolling back the terminal, that's how I copied the md5sums of a stale md5sums file - that was even already wiped BTW :-(

[Carnë Draug]

You are placing limitations on the package requirements based on what you think will happen in the future. But what you think will happen is based on very little and you have already been wrong. The end result are package versions that won't install on recent versions of Octave, even though they would work just fine.

Anyway, you are obviously conscious of your decision and it's your call. It doesn't matter how much I disagree. I have pushed the release.