Hello, again.
I would like to know is there a way to see if this proposal rejected or
something? Is this a write ML to propose with patch?
And yes I saw <ddevie...@gmail.com> suggested to use checksums, but this
means we read all the files in a source directory, that is huge. This is
a reason for relying on timestamps. If some os does not update its files
timestumps why we suffer.
On Thu, 21 Mar 2019, 10:31 Захар Малиновский, <zakhar.malinovs...@gmail.com>
wrote:
> Yes, timestamps only.
> Sqlar does not has a column for checksum or sha.
>
> Here comes the patch:
>
> Index: src/shell.c.in
>
> ==================================================================
>
> --- src/shell.c.in
>
> +++ src/shell.c.in
>
> @@ -5820,10 +5820,30 @@
>
> " mtime,\n"
>
> " data\n"
>
> " FROM fsdir(%Q,%Q)\n"
>
> " WHERE lsmode(mode) NOT LIKE '?%%';"
>
> };
>
> +
>
> + // updates only newer files
>
> + const char * zInsertFmtNewer =
>
> + "REPLACE INTO %s(name,mode,mtime,sz,data)\n"
>
> + " SELECT\n"
>
> + " %s,\n"
>
> + " fs.mode,\n"
>
> + " fs.mtime,\n"
>
> + " CASE substr(lsmode(fs.mode),1,1)\n"
>
> + " WHEN '-' THEN length(fs.data)\n"
>
> + " WHEN 'd' THEN 0\n"
>
> + " ELSE -1 END,\n"
>
> + " sqlar_compress(fs.data)\n"
>
> + " FROM fsdir(%Q,%Q) as fs\n"
>
> + " left join sqlar on sqlar.name = fs.name \n"
>
> + " WHERE lsmode(fs.mode) NOT LIKE '?%%' \n"
>
> + " and (fs.mtime > sqlar.mtime or sqlar.name isnull) \n"
>
> + " ;\n";
>
> +
>
> +
>
> int i; /* For iterating through azFile[] */
>
> int rc; /* Return code */
>
> const char *zTab = 0; /* SQL table into which to insert */
>
> char *zSql;
>
> char zTemp[50];
>
> @@ -5856,13 +5876,23 @@
>
> if( rc!=SQLITE_OK ) goto end_ar_transaction;
>
> }
>
> rc = arExecSql(pAr, zCreate);
>
> }
>
> for(i=0; i<pAr->nArg && rc==SQLITE_OK; i++){
>
> - char *zSql2 = sqlite3_mprintf(zInsertFmt[pAr->bZip], zTab,
>
> - pAr->bVerbose ? "shell_putsnl(name)" : "name",
>
> +
>
> + char *zSql2;
>
> + if( bUpdate!=0 && !pAr->bZip) {
>
> + // updates only newer files
>
> + zSql2 = sqlite3_mprintf(zInsertFmtNewer, zTab,
>
> + pAr->bVerbose ? "shell_putsnl(fs.name)" : "fs.name",
>
> pAr->azArg[i], pAr->zDir);
>
> + } else {
>
> + zSql2 = sqlite3_mprintf(zInsertFmt[pAr->bZip], zTab,
>
> + pAr->bVerbose ? "shell_putsnl(name)" : "name",
>
> + pAr->azArg[i], pAr->zDir);
>
> +
>
> + }
>
> rc = arExecSql(pAr, zSql2);
>
> sqlite3_free(zSql2);
>
> }
>
> end_ar_transaction:
>
> if( rc!=SQLITE_OK ){
>
>
>
> On Wed, 20 Mar 2019, 18:18 Dominique Devienne, <ddevie...@gmail.com>
> wrote:
>
>> On Wed, Mar 20, 2019 at 5:13 PM Захар Малиновский <
>> zakhar.malinovs...@gmail.com> wrote:
>>
>> > Here I include patch file with the changes: fossil diff --from trunk
>> .\src\
>> > shell.c.in > patch-updateOnlyNewer.patch
>> > (see patch file patch-updateOnlyNewer.patch)
>> >
>>
>> Copy/paste it inline to your message. Attachments are stripped on this ML.
>> --DD
>>
>> PS: based on timestamps only? Using a checksum (MD5, SHA1, SHA3, whatever)
>> would be "safer", if a little slower.
>> Of course, that assumes SQLar uses a checksum. I don't recall, I thought
>> not.
>> _______________________________________________
>> sqlite-users mailing list
>> sqlite-users@mailinglists.sqlite.org
>> http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
>>
>
_______________________________________________
sqlite-users mailing list
sqlite-users@mailinglists.sqlite.org
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
