From: Jeff Hostetler <git@jeffhostetler.com>
Date: Fri, 22 Mar 2019 14:20:02 GMT
Subject: Re: [PATCH v2 1/1] trace2: write to directory targets
Message-ID: <b0d5949a-1afa-fcce-646c-d33d51686a2f@jeffhostetler.com>
In-Reply-To: <xmqqpnqjr3g0.fsf@gitster-ct.c.googlers.com>



On 3/21/2019 11:30 PM, Junio C Hamano wrote:
> Jeff Hostetler <git@jeffhostetler.com> writes:
> 
>>>> +	for (attempt_count = 0; attempt_count < MAX_RANDOM_ATTEMPTS; attempt_count++) {
>>>> +		strbuf_reset(&final_path);
>>>> +		strbuf_addbuf(&final_path, &base_path);
>>>> +		strbuf_addf(&final_path, ".%d", attempt_count);
>>
>> Since the last component of the SID is already very unique and
>> we're unlikely to have collisions, maybe change the above line to:
>>
>> 		if (attempt_count > 0)
>> 			strbuf_addf(&final_path, ".%d", attempt_count);
>>
>> and in reality expect to never have files with the suffix.
> 
> That's a nice property.
> 
>> Unless, that is, they turned on more than one of GIT_TR2,
>> GIT_TR2_PERF, or GIT_TR2_EVENT and pointed them at the same
>> directory, but I'm not sure if I care about that edge case
>> or not.
> 
> That actually makes me wonder if the auto generated filenames want
> to have a common trait (e.g. suffix) that allows the readers to tell
> from which of these environment variables the names came from.  It
> would not be very useful if two files with the same sid component
> had .1 suffix for GIT_TR2 trace for one session, and the same .1
> suffix is used for GIT_TR2_PERF trace for a pair of files from
> another session.

I thought about suggesting that, but didn't think it worth the bother.

> But let's not worry about it for now.  If people do not want them
> get mixed up and become hard to tell apart, they can always specify
> different directories for different traces.

agreed. it should be very rare.  the SID is the built from
	"<microseconds-since-the-epoch><dash><pid>"
so something would have to be seriously wrong with their
system to get collision from 2 different git commands.

and yes, they the advise should be to use different directories
for the different streams.

> 
>>
>>>> +		fd = open(final_path.buf, O_WRONLY | O_CREAT | O_EXCL, 0666);
>> [...]
>>
>> Nice.  Thanks for looking into this.
>> Jeff