From: Jeff Hostetler <git@jeffhostetler.com>
Date: Tue, 26 Jul 2016 13:53:09 GMT
Subject: Re: [PATCH v2 5/8] status: print per-file porcelain v2 status data
Message-ID: <57976B45.9060102@jeffhostetler.com>
In-Reply-To: <xmqqwpk94gbu.fsf@gitster.mtv.corp.google.com>



On 07/25/2016 04:23 PM, Junio C Hamano wrote:
> Jeff Hostetler <jeffhost@microsoft.com> writes:
>
>> +static void wt_porcelain_v2_print(struct wt_status *s);
>> +
>
> There is no point in this forward declaration, if you just place the
> implementation of these functions here, no?

Right. I just did it that way to make the diffs with the previous
commit draw a little cleaner.  But I can take it out.


>> +/*
>> + * Print porcelain v2 info for tracked entries with changes.
>> + */
>> +static void wt_porcelain_v2_print_changed_entry(
>> +	struct string_list_item *it,
>> +	struct wt_status *s)
>> +{
>> +...
>> +	fprintf(s->fp, "%c %s %s %06o %06o %06o %s %s R%d %s",
>
> It is misleading to always say R in the output when there is no
> rename, isn't it?

Yes, especially if we add it for copied entries too.
I was just looking for a way to have a fixed format.
If we make the R%d field optional, then the first pathname
is ambiguous.  That gets me back to an earlier draft where
we have rename and non-rename line types.

I'll split this up into 1 pathname and 2 pathname forms,
and only include the R%d (or the C%d) field in the latter.

>
>> +	 * Note that this is a last-one-wins for each the individual
>> +	 * stage [123] columns in the event of multiple cache rows
>> +	 * for a stage.
>
> Just FYI, the usual lingo we use for that is "multiple cache entries
> for the same stage", I would think.

thanks.

>
>> +	 */
>> +	memset(stages, 0, sizeof(stages));
>> +	sum = 0;
>> +	pos = cache_name_pos(it->string, strlen(it->string));
>> +	assert(pos < 0);
>> +	pos = -pos-1;
>> +	while (pos < active_nr) {
>> +		ce = active_cache[pos++];
>> +		stage = ce_stage(ce);
>> +		if (strcmp(ce->name, it->string) || !stage)
>> +			break;
>> +		stages[stage - 1].mode = ce->ce_mode;
>> +		hashcpy(stages[stage - 1].oid.hash, ce->sha1);
>> +		sum++;
>> +	}
>> +	if (!sum)
>> +		die("BUG: unmerged entry without any stages");
>
> Hmm, we seem to already have d->stagemask; if you call that variable
> "sum" anyway, perhaps its computation can be more like
>
> 	sum |= 1 << (stage - 1);
>
> so that you can compare it with d->stagemask for this sanity check?

good point.

thanks
jeff