Without knowing the DS1307 well, I think there are two options:
Make the setting of the clock a UI function (you should probably do this in any case since you will want users to have the ability to set the clock). The user would have to initiate the action (and put in the time to set) so there would be no problems the clock getting reset inappropriately, or
Do a check for sanity of the clock before you run that line. You might be able to do that through the clock API (e.g., make a call to check the time, or to get the clock status and see if it makes sense) or you could do it by storing the clock status in nonvolatile memory. The very worst case would be that you store the time of last boot and check that it seems reasonable (e.g., the time you get from the RTC is after the last boot and is within reason – say no more than a month or a year ago). If the time is too far off to feel confident you could issue a "check time" warning.
If you have a network connection using NTP as well as the RTC would be a very smart option. NTP can hold the time within a few milliseconds. The RTC would then provide you with protection against a network outage.
If you are using the Arduino to manage scheduled events one thing you'll want to think about is events that are scheduled in a moment of time that is either "skipped" as you advance the clock or that "repeats" if you set the clock back. One way to handle that is to never "jump" time, but to always just speed it up or slow it down.
Without knowing the DS1307 well, I think there are two options:
Make the setting of the clock a UI function (you should probably do this in any case since you will want users to have the ability to set the clock). The user would have to initiate the action (and put in the time to set) so there would be no problems the clock getting reset inappropriately, or
Do a check for sanity of the clock before you run that line. You might be able to do that through the clock API (e.g., make a call to check the time, or to get the clock status and see if it makes sense) or you could do it by storing the clock status in nonvolatile memory. The very worst case would be that you store the time of last boot and check that it seems reasonable (e.g., the time you get from the RTC is after the last boot and is within reason – say no more than a month or a year ago). If the time is too far off to feel confident you could issue a "check time" warning.
Without knowing the DS1307 well, I think there are two options:
Make the setting of the clock a UI function (you should probably do this in any case since you will want users to have the ability to set the clock). The user would have to initiate the action (and put in the time to set) so there would be no problems the clock getting reset inappropriately, or
Do a check for sanity of the clock before you run that line. You might be able to do that through the clock API (e.g., make a call to check the time, or to get the clock status and see if it makes sense) or you could do it by storing the clock status in nonvolatile memory. The very worst case would be that you store the time of last boot and check that it seems reasonable (e.g., the time you get from the RTC is after the last boot and is within reason – say no more than a month or a year ago). If the time is too far off to feel confident you could issue a "check time" warning.
If you have a network connection using NTP as well as the RTC would be a very smart option. NTP can hold the time within a few milliseconds. The RTC would then provide you with protection against a network outage.
If you are using the Arduino to manage scheduled events one thing you'll want to think about is events that are scheduled in a moment of time that is either "skipped" as you advance the clock or that "repeats" if you set the clock back. One way to handle that is to never "jump" time, but to always just speed it up or slow it down.
Without knowing the DS1307 well, I think there are two options:
Make the setting of the clock a UI function (you should probably do this in any case since you will want users to have the ability to set the clock). The user would have to initiate the action (and put in the time to set) so there would be no problems the clock getting reset inappropriately, or
Do a check for sanity of the clock before you run that line. You might be able to do that through the clock API (e.g., make a call to check the time, or to get the clock status and see if it makes sense) or you could do it by storing the clock status in nonvolatile memory. The very worst case would be that you store the time of last boot and check that it seems reasonable (e.g., the time you get from the RTC is after the last boot and is within reason – say no more than a month or a year ago). If the time is too far off to feel confident you could issue a "check time" warning.