update version number in example to 0.3.5

[gubatron]

No description provided.

[jbenet]

@gubatron as a user, is this disconcerting?

i don't want to have to update this every single time we push a new version. we push versions very often and this would add annoying friction. If it was automated, that could work, but deployment to ipfs is not yet automated.

[gubatron]

let's make it automated, this is indeed annoying.
is there a single point (maybe a changelog file) that we could parse? (maybe an ajax request somewhere)

As a user, this gave me the impression that the page hasn't been updated in a long time and that whatever else I was reading might not be true to how the setup process was.

[jbenet]

As a user, this gave me the impression that the page hasn't been updated in a long time

it hasn't needed to be updated.

and that whatever else I was reading might not be true to how the setup process was.

It must still be. otherwise it is an error to be corrected.

We endeavor to keep that page updated, however in the event there is a discrepancy, a fake version could be doubly confusing to the user. If the listing says "version 5.6.7" then the user would expect everything else to have been generated with "version 5.6.7", which would not be the case. It may make finding a problem harder.

What about including a note as to the version mismatch between what's listed on the page and what the user would be seeing?

[gubatron]

I think the note would be lame, might as well just put x.y.z then.

I also see that some pages seem to work with a templating system of some sort (no clue how the site actually works), so in theory, if this install page were to work using a templating system, the latest version number could be pulled from somewhere into a variable available sitewide. Ideally this variable would be saved automatically to a version file somewhere where it could be loaded to every template that would need it as part of your software release cycle, as you must change the version number at some point, and ideally you'd do this only in a single place #DRYPrinciple

[jbenet]

I also see that some pages seem to work with a templating system ...

Of course, reading a current version from the repo is not the problem.

The problem is misleading the user by pretending to have generated the page with the version they have instead of an earlier version. If there is indeed a discrepancy in the examples it would be very disconcerting and make it seem something is broken in the client instead of broken in the example page. If there is a different version number, at the very least the user can realize that perhaps something could have changed.

Think about like this, man pages usually carry versions and dates, to make it clear when the man page was written and that if the version of the tool does not match the man page then things may be different. if man pages just used tool -v and date then it would be very, very confusing if the man page mismatched what the user was observing.

[jbenet]

One thing we could automate is a proper update that validates the examples. Instead of just pulling the latest version, a script would run all the examples and verify they still generate the same output. Only then is the version number safe to be updated.

[jbenet]

regardless, thanks for bringing this to our attention @gubatron