Is a Project Save 100% safe?

[Mark_Strickland]

I have a project that has grown pretty large and takes a few seconds to save even on my MacBook with 3.1Ghz i7 processor and SSD. Some days there is a fair amount of development activity and if the project was corrupted during the day going back to the previous backup the day before would be “painful”.

Does Xojo somehow protect against project corruptions from an IDE or system crash on a save by NOT overwriting the original file until it knows the save was successful?

I have never had a problem but just thinking about what could happen. I don’t run TimeMachine continuously but I suppose that is an option.

[Norman_P]

Use Version control and never worry about this again
And use an offsite service like bitbucket, assembla’s svn hosting etc in conjunction with time machine locally then you dont even have to worry about what happens if your house suffers a flood, fire etc

[Tim_Parnell]

+1 Definitely use version control.
Any doubts? Ask Dave how it goes when something goes wrong and suddenly everything is dated 2014

[Norman_P]

yes… it saves to a “temp” file… when successful, it deletes the original and renames the temp.

No
It uses OS facilities to do atomic swaps of files
Which is far safer

it also seems to do incremental saves along the way… as I have had times when the computer froze, I rebooted, and Xojo said “I found unsaved changes, would you like to load them?” (words to that effect)… and when I did, I was right back where I’d been before the freeze.

No
It save a temporary copy whenever you run
Not incremental changes

[quote]Is it perfect? will you never possibly lose something? Nah… but proper practices (save often) and you should be good to go… While I am not a fan of VCS systems, I AM a fan of having some kind of back up method.
[/quote]
VCS systems are a life saver - not only because you can roll back to previous versions and find which change introduced a bug (which is immensely handy) but because you can branch , work on new features in that branch and then later merge all those change back in quite simply
Doing this by hand for a team of more than 1 is a serious pain in the ■■■■
Been there done that on a fairly decent sized project BEFORE there was the VCP format - immensely painful
And do not want to repeat that

[Norman_P]

here is NO WAY in H*LL I would ever use a VCS system… unless they now save WHOLE unaltered copies of the project

Some do

(which I doubt has change)… Every VCS system I was “forced” to use in the past, utilized a DIFF engine and stored only what was necessary to “patch” an older version… And on more than one occasion, those patches fouled up and the code that came out was NOT the code that went in…

never had that issue in 30+ years of using them in various forms on different platforms
From DEC VMS versioned files (perhaps the simplest and easiest to use since it was built right into the OS) CVS SVN for 15+ years and even some git projects
Obviously your experience varies but in this day & age they are so common that any place anyone wants to work will use one

Individual developers manually doing version control is just masochistic

And if newer VCS systems DO store complete code base, then they waste space… and no matter what they will exceed a “few seconds” to save…

takes no longer to check in changes to my local SVN than it does to save a project
in fact with small changes its faster

And in the scenario that Tim mentioned, VCS would not have helped… because as it turned out, I had made the changes, compiled the changes, and then neglected to SAVE the changes… VCS is NOT magic…

No its not - it’s an ■■■ saver ( as in CYA)

So as I have said before… on this subject you are all free to agree to disagree with me.

vehemently

And Norman… while the mechanics I described may not be “exactly” what happens behind the scenes, the end result is the same… and yeah my use of the term “incremental” was incorrect, “temp copy” was more what I meant

No its not - and understanding why its not is important
Atomic swaps are not the same as “delete original and rename”
Thats why we use them
Delete + rename can fail if you get a power outage / hardware failure right as the file has been deleted & the other NOT yet renamed

Atomic swaps either completely succeed or fail - even in the case of a power outage or hardware failure
Its why OSes provide those facilities

[Tim_Parnell]

Dave I realize you don’t trust vcs and I’m not trying to change that. I was just trying to make light of the situation and encourage others to try a version control system.

I apologize for my remark.