unfinished documents:
pfe sources pfe manpages pfe docbook |
2003-02-19 "0.7.18" <guidod@gmx.de> * taking over make-dbk.pl from zziplib, place pfe-intro.dbk into doc/pfe, and generate pfe-forth.dbk and pfe-forth.html from these sources. This makes for a perfect docbook master. * `make pfe` will now create doc/pfe/pfe-forth.html 2003-02-18 "0.7.17" <guidod@gmx.de> * making "$id" part of non-fcode entries to be just secondary to the $forth_name: makes for single-digit serial numbers. :-) * s/xml_text_*/xml_strstr_*/ * s/xml_node_append_new/xml_node_append_node_new/ * s/xml_node_append_new_data/xml_node_append_node_new_data/ * s/xml_node_append_text/xml_node_append/ * enhance xml_node_append with <*%p/ = attribute_list_copy support to make formatted tree_append more powerful - then go to pdoc-pfe and replace these parts which should enhance the readibility and maintainability of that code part considerably. 2003-02-17 "0.7.16" <guidod@gmx.de> * add warning_default_pkgconfig * defaulttext for synonyms * oops - where is the export_value to attribute detection... * introducing vararg append_text to appends nodes and text snippets at the same time using a compact syntax. 2003-02-14 "0.7.15" <guidod@gmx.de> * change makefile for "doc" files - to make them all somewhat look like html and bind them together with a common navbar into a complete website 2003-02-13 "0.7.14" <guidod@gmx.de> * complete <link> detection - all "=>" parts are not detected and largely resolved as well. * xml_pdoc_c_doc reimplemented - does now properly recognize alternating multi-line sections of <para> and <screen>. * cleanup cblocks messing up doc section 2003-02-13 "0.7.13" <guidod@exgate.tek.com> * implement <link> detection and link-resolution * implement indexTerm generation - the microsoft htmlhelp compiler can not stand it, it has a fixed maximum of 1500 entries. 2003-02-12 "0.7.12" <guidod@gmx.de> * implement the speedup-variant for the report generation using an index-list for the cblocks we want to merge into the output (... it happens to be too fast to implement sth even quicker) * fcodename gets to know again about number-to-name conversion * docbook generation - after lots and lots of trial and error, I found a layout that gets rendered nice for both html and troff 2003-02-11 "0.7.11" <guidod@exgate.tek.com> * expand configure/make to allow binding of extra forth sources from the mforth source tree * make-max will then build the complete words.xml / wordsets.xml file and the words.docbook of course : 1179 refentry parts 2003-02-11 "0.7.10" <guidod@gmx.de> * merge tek tree with pub tree * output words.xml integrated 2003-02-10 "0.7.9" <guidod@gmx.de> * xml_pdoc_pfe_wordsets_2_words_reference * xml_node_remove needed * xml_node_foreach needs a fix: what if the "node" gets removed? 2003-02-09 "0.7.8" <guidod@gmx.de> * create doc/pfe/wordsets2words.pl * xmlto html-nochunks of the result 2003-01-11 "0.7.7" <guidod@gmx.de> * change every instance of GList* into using xml_GList* instead * make xml_GList union s/item/data/ and the enclosed gpointer gets s/data/pointer/ * bin/xml-c-pfe-doc to print the report format (wordset/exports/doc) * actual changes to support pfe report generation * try `make wordsets.html` to see 2003-01-10 "0.7.6" <guidod@gmx.de> * the readability got lost - let's change the naming scheme :-) s/xml_path_nodes_/xml_path_pcre_/ s/xml_tree_nodes_/xml_path_node_/ * renaming header files (oops) basically inverting the naming scheme. s/nodepath./pathnode./ s/pcrepath./pathpcre./ s/nodeadds./addsnode./ s/pcreadds./addspcre./ s/nodeattr./attrnode./ s/nodecopy./copynode./ s/nodedata./from-xml./ s/pcredata./from-xmf./ s/nodedump./dumpnode./ s/pcredump./dumptree./ s/nodefile./from-txt./ s/nodelist./listnode./ s/pcrelist./listpcre./ s/nodenext./nextnode./ s/pcrenext./nextpcre./ s/nodesave./savenode./ s/pcresave./savetree./ s/pcrenode./greptree./ 2003-01-02 "0.7.6" <guidod@gmx.de> * trigger also at "." since it occurs quite often as firstchar in regex things - at the same time it is not used for a key value in xpath syntax. 2002-12-31 "0.7.6" <guidod@gmx.de> * s/xml_g_strstr/xml_g_strstr_len/ * introduce xml_g_strstr with 3 arguments - and make use of it. * regex only triggered if strchr("^(?+|\\", firstchar) atleast now in xml_path_nodes_list/foreach functions * note that neither "+key" and "^key" must match up to the end. * xml_path_nodes_add: s/xml_tree_nodes_/xml_path_nodes_/ 2002-12-30 "0.7.5" <guidod@gmx.de> * move aux files into use/ subdir * remove GNU/NOGNU fuzz about cd-and-build gmake does it right and notices the "cd" as if calling "-C" * s/xml_path_nodes_/xml_tree_nodes_/ and want to de-write it into routines that do not understand xpath separators - they will only check for a single element of an xpath. * s/xml_pcre_nodes_/xml_path_nodes_/ and will make it the only routine to understand xpath syntax - the pcre mode is only triggered when asked for with "(" as described now. * s/xml_path_text_to_attribute/xml_tree_nodes_text_to_attr/ * s/xml_path_attribute_to_text/xml_tree_nodes_text_restore/ * s/xml_pcre_text_to_attribute/xml_path_nodes_text_to_attr/ * s/xml_pcre_attribute_to_text/xml_path_nodes_text_restore/ * hmmm, okay, need to decide it differently: A = strcmp match * = strstr match ( = pcre match complete (makes extra match on complete) ^ = pcre match (implicitly) anchored subportion ? = pcre match complete + = pcre match subportion | = pcre match complete also-empty ... and need to change the `make check` tests as well, with this specifications however it can be turned into the new world by adding a "?" in front - so that "//K+" does simply become "//?K+" to trigger a pcre complete, and the current pcre "//*K" must be written now as "//+K" to match again. * for the current tarball however, we prepare it by just adding (!!) "+" notification to all the pcre-routines that do currently interpret "*" and then we move over the current pcre calls and: (a) add "?" or (b) make "*" -> "+". * and last not least, add scrollkeeper registration. 2002-12-10 "...." * adding TEXTMINING doc to provide links to projects in a nearby domain of text analysis, text retrievial and text visualization. 2002-12-07 "...." * reading up with the xpath2 specification, I do derive a behaviour change to be interesting. The current node_match functions of strstr/strcmp style and prce/anchored are collapsed. The strstr/strcmp style behaviour survives largely, but pcre gets only enabled if the first character of a match-part happens to be a "(" subpattern character. The user is free to make that a "(?:" to not get listed in the ovector pattern return table. * this will likely speedup operations as well: on an xpath foreach/list detection, the usual parts will be matched as strcmp/strstr style while only specific subportions will be compiled as pcre and checked on a loop. * furthermore it makes maintainance easier since only one group of foreach/list routines exist. However, lot's of example code needs to be changed in consequence as well. * The "(" detection makes it also easier to derive the end of a pcre pattern by looking for the matching close-")" paren. That makes pattern detection more save and simply intuitive to the user. * While being there, '".."' subportions of matches might need to be checked as well - that's for @attrib="value" checks. Shall we also mandate that @attrib=value means '=~' ?? hmm. * All these changes have to wait up until next thursday however. * Summary: check the first char of a node-match: case "(": compile as /x pcre and match the node-name or attrib-name - as anchored match (as default). case '?': shorthand for not-anchored not-empty pcre case '\\' shorthand starting an anchored pcre /x pcre case "*": use strstr on the node-name/attrib-name default: use strcmp on the node-name/attrib-name 2002-12-05 "..." <guidod@gmx.de> * expanded XPATHDEF with comparisons on xpath selections and it postfix/prefix expressions. 2002-12-04 "xmlg 0.7.4" <guidod@gmx.de> * nodeattr.h - remove usage of xml_node_attribute_foreach from any of the lib files. * pdoc-cpp - put "-J" as "J text='saved'" markups, that's better visually in highlighted text. It still removes the "*"-star but leaves whitespace thanks to xml_node_save_text_to_attribute * nodeattr.c - remove all xml_node_attribute_foreach* functions * insterestingly the markup scripts do now run *faster*, it seems the overhead for the indirect-call in the foreach functions was not liked on pentium. And the grep_attributes does now have the chance to "break" on the first hit, another speedup. 2002-12-03 "xmlg 0.7.3" <guidod@gmx.de> * nodeattr.h missing in xml/Makefile.am, so it got not tarred. * cleanups w.r.t. pre-C99 compilers * some ln -s needs to be cleaned up * a number of build cleanups * adding doc/pfe - and have special handling for the in-project example * pdoc-pfe gets expanded with export-name as a prerequisite for making a pfe wordset report * doc/pfe/pfe.xml via bin/xml-c-pfe-cat to combine all sources into one big master file. * printing the master file does not work - is that an OOM ? * printing the pfe.xml does work now - it was a bug in the dump routine where the case of an null.text root node was not handled correctly, it is now. btw, the pfe master xml from 1MB sources will be 5MB in full xml, and converting it to css html makes it 10MB. However, those conversions range in 1..10 seconds - the xm-tool.sf.net did require about 30..50 minutes - about 100 times longer!! 2002-12-01 "xmlg 0.7.2" <guidod@gmx.de> * xml_node_attributes_copy(node) -> xml_attr_list_copy(node->attribs) * GNode->attributes becomes an xml_AttrList * severely shorten nodeattr.* by reusing attrlist.* functions and get rid of its _G/_L casts which are not needed anymore * it seems to be working, hooray! err, as different to 7.0 excercise 2002-12-01 "xmlg 0.7.1" <guidod@gmx.de> * attrlist.h added - that's more special than glist.h * attrdata.h takes the common part of attrlist.c and nodeattr.c * nodeattr.c does now include attrdata.h and gets converted atleast for using xml_attr_data_new/free 2002-12-01 "xmlg 0.7.0" <guidod@gmx.de> * glist.h added - not very useful up to here. * rename xml_g_node_attribute_* into xml_node_attribute_* * changing over to xml_GList* attributes did not work however, I am too tired of the day obviously... switched back and tarred what has been there so far. 2002-11-30 "xmlg 0.6.9" <guidod@gmx.de> * doc added: NODEMODEL and XPATHDEF * doc dir added: doc/bin to highlit bin examples * fix various packaging problems 2002-11-29 "xmlg 0.6.7/8" <guidod@gmx.de> * xml_pcss-scanning did not work on missing ";" at end of proplist * nodenext gets helper routines xml_test_name_ and friends that shall replace ALL instances of strcmp/strstr currently around. * rename-tree goes through helper functions to differentiate between html and xml output. * doc/xml subdirectory to take html'ized sourcecode - actually a realworld test for the current css-to scripts. And an example of what a simple hypercss transformation can look like. NOTE: the current order of SUBDIR evaluation is wrong, the toplevel executables should be moved into their own subdir but that is left for the next release. * xml_node_group_outer_alias for nodeaddds.*, as we want it for * html-func to intelligently set the "mark" and also add outer markup to nodes, both depending on the "display" attribute. * sereious bug in _cblock-markup routine - obviously the following C construct does not (!!) loop - it might be a bug, but, well.. do { gchar* p = memchar(...); if (p) continue; } while (0); 2002-11-27 "xmlg 0.6.6" <guidod@gmx.de> * xml-css-tree test-program to show the result from * xml_pcss_prop_list_from which makes a SELECT copy of prop_head and its prop_value's into its own xml list. * nodecopy has now a test routine (however, not the new shadow_copy routines, just the append_copy routines). * xml_pcss_rename_proc does actually work now * xml-css-to does actually work now in consequence, too, that is that it does rename elements according to a css * ee.css as an example and `make check-rename` to show you to use it. It's not included in an automatic check however. Note how "C" elements are transformed but the latest "C" is different for being in an when-inside rule. * ... do we need to sort by "specificity" of css rules? 2002-11-27 "xmlg 0.6.5" <guidod@gmx.de> * nodeattr - oops, sorting was in wrong order, I did change that during debugging last time, and did not reverse it. * check - routines need to be adapted to new order, of course * have.a.look at `./xml-css-read e.css`, the sorting of attributes makes things look better. 2002-11-27 "xmlg 0.6.4" <guidod@gmx.de> * xml_attr becomes its own type - and gets its own header file basically, we change the implementation type from a g_hash_table<str,str> into g_list<attr> and attr is a struct mostly not to be used outside. * to.clarify - most nodes do not even have a single attribute and then just one and so on. we just keep a sorted GList for these - and it make it rightly easier to walk and join these. * check-routines should only be dependent of the binaries they are about to use in their test - that is possible since we have each of it with the .la dependency and a rule that can recures into xml/. for make those. * oops - since we have alphabetic order, the .out mgiht sometimes just be different 2002-11-27 "xmlg 0.6.3" <guidod@gmx.de> * _hard_life_ - work interrupted by the death of my father, he died unexpected from an silent heart attack. Stupid laws do not allow to pull the plug from the breathing machine to let him die the way that humans die. That's because most of YOU have not yet been thinking about the possilibility that it might be YOU to be in his place and being held back on earth for hours instead of being allowed to enter the light. Give it a minute! * index.html - generated automatically with content of README * README - wording slightly corrected but still not up to date. * xml/nodecopy.* added to own functions for copying/renaming * xml/node*.h - fix once-def which is still XML_G_* someplace * xml/nodeadd - add "=attrib" syntax to add9 - needed for the css selector scanning. Furthermore, we add multi attribute insertion on "=" separator. * xml/pcss-src - scan selector syntax to prop_head attributes ... here you can already see the class<->mark relation. * e.css - for testing ./xml-css-read and see what I mean... 2002-11-23 "xmlg 0.6.3" <guidod@gmx.de> * HYPERCSS document added * xml-css-to wrapper added * xml/pcss-proc.* added to own the work functions for hCSS * xml/pcss-src needs to skip "<!--" at the start of a css spec 2002-11-21 "xmlg 0.6.2" <guidod@gmx.de> * create pcss-src.* files and routines along with a test prog named xml-css-read. * pcresave needs the ability to specify the fillchar * pdoc-cpp use another fillchar and correct _cblocks routine * css scanning is brought to a state of having nodes with <prop_value prop="name" media="media">..</prop_value> 2002-11-21 "xmlg 0.6.1" <guidod@gmx.de> * create libxmlglib.pc and libxmlpcre.pc files for lib/pkgconfig * install also xml/xmlglib.h and xml/xmlpcre.h header files autogenerated from library header information for automake * rename pcresrcs pcrecdoc and pcrecpfe into pdoc-src pdoc-cpp and pdoc-pfe and adapt the function names from xml_pcre_ into xml_pdoc_ - however leave these sources as part of libxmlpcre as they do not fatten that lib all too heavily * update xmlg.spec file accordingly (forgot includes there) 2002-11-20 "xmlg 0.6.0" <guidod@gmx.de> * rename xml/xmlg* into xml/node* which clarifies things and makes it more readable. * rename a lot of routines named directly as "xml_g_<method>_" into "xml_tree_" for clarification. Everything that is not particularly a helper routine for node-handling gets renamed from xml_g_node to xml_node. The xml_g_node_*'s are now only for juggling node fields at the toplevel. s,xml_g_to_,xml_tree_to_,g s,xml_g_find_,xml_tree_find_,g s,xml_g_text_,xml_node_text_,g s,xml_g_dump_,xml_tree_dump_,g s,xml_g_group_outer,xml_node_group_outer,g s,xml_g_group_inner,xml_node_group_inner,g s,xml_g_node_to_,xml_tree_node_to_,g s,xml_g_node_inside,xml_tree_node_inside,g s,xml_g_node_match,xml_node_match,g s,xml_g_node_contains,xml_node_contains,g s,xml_g_node_followedby,xml_node_followedby,g s,xml_g_node_hasnextnode,xml_node_hasnextnode,g s,xml_g_node_prev_text_,xml_node_prev_text_,g s,xml_g_node_next_text_,xml_node_next_text_,g s,xml_g_node_this_text_,xml_node_this_text_,g s,xml_g_node_next_name,xml_node_next_match,g s,xml_g_node_next_or_last,xml_node_next_or_last,g s,xml_g_node_copy_,xml_node_copy_,g s,xml_g_node_find,xml_node_find,g s,xml_g_node_grep,xml_node_grep,g s,xml_g_node_save_,xml_node_save_,g s,xml_g_add,xml_tree_add,g s,xml_g_append_new,xml_node_add_new,g s,xml_g_appends_new,xml_node_adds_new,g s,xml_g_group_cut,xml_node_group_cut,g * rename and modify node_parse routines - they expect a gstring as an argument, and only if that is left null then a new one is allocated - otherwise the parsed text will be added to the end of that gstring and alls offs of nodes made related to that gstring. xml_g_node_parse xml_g_markup_parse_text xml_g_node_parse_file xml_g_markup_parse_file xml_pcre_node_parse xml_g_parse_text xml_pcre_node_parse_file xml_g_parse_file xml_g_node_text_file xml_tree_node_from_file xml_tree_node_from_text * gerror - in the process, add gerror helpers to the main programs where noticing they are not used yet. * nodenext - add macros that are renames of the str-check funcs as xml_node_match_eq_ == !strcmp and xml_node_match_as_ == strstr and go through some files placing these quick ones where possible to speed up things even more than the lib itself does by design. ... and do a similar thing for the two xml_node_hasnextnode_as_ and xml_node_hasnextnode_eq_. * nodenext add the two xml_node_not_match_eq_ and not_match_as_ which are capabable of checking arg node for null. And some quick ones for xml_node_notnextnode_... * ...next ... arg, shut up, and just rename all those calls from xml_node_match* to xml_node_hasname* and of course node_not_match being node_notname ... and node_next_match as node_next_with_name * and now make xml_node_text_empty/match into what it should have been in the first place: xml_text_empty and xml_text_match and do not forget xml_pcre_match.. * corrige <br>'s in commentblocks - and make item_bloc the default for pfe-read routine in order to get at nice line-info in the item markups 2002-11-19 "xmlg 0.5.9" <guidod@gmx.de> * did work a a lot on actual markup routines examing the text and markup tree - folded into pcrecdoc and a newer cousin pcrecpfe. In the course some helper routines were added. 2002-11-18 "xmlg 0.5.7" <guidod@gmx.de> * pcreadds - rename xml_pcre_add_ functions to be named xml_pcre_nodes_add_ since they actually make a path selection on nodes instead of handing over a specific node. * xmlgnext - remake xml_g_name_match0 (RE, text, off, end) into xml_g_text_match0 (text, off, end, RE). The same for pcrenext. That shall be the rule - instead of g_node with one leading arg it is now g_text with three args being in up front on the call stack. * stroustrup - walk through all library texts and add an emacs footer for c-file-style: "stroustrup". * pcrenext - add xml_pcre_text_match_add9 which does hand over the match-vector to xml_g_add9 for markups. 2002-11-12 "xmlg 0.5.6" <guidod@gmx.de> * pcrecdoc - adding more routines - the C text is now even separated into itemcdoc/itemcdef sections which are the ones we want to scan later for information. 2002-11-12 "xmlg 0.5.5" <guidod@gmx.de> * xmlgdump.*/pcredump.* - need to spit out entities for xml reserved chars. Use helper called xml_g_string_append in the place of g_string_append_len (g, node->text->str...) * pcredump.*...etc - rename xml_g_to_xxml into a word named xml_g_dump_to_xml since it can also handle xml_g_dump as an output. And it reads better. Likewise we rename the old xml_g_to_xml into xml_g_node_to_xml to clarify things. * xml/gerror.* - make printf-format for xml_g_show_error * fix.exmaple.programs using the new techniques 2002-11-12 "xmlg 0.5.4" <guidod@gmx.de> * configure.ac AC_PREFIX_CONFIG_H to AX_PREFIX_CONFIG_H * rename some API entries, do not call it xml_g_xpath but now xml_path_, actually kill the use of _xpath_* completly in the sources, and try to make it path_nodes or pcre_nodes xml_g_xpath_* -> xml_path_nodes_* xml_pcre_xpath_* -> xml_pcre_nodes_* xml_g_xpath_attribute_to_text -> xml_path_attribute_to_text xml_g_xpath_text_to_attribute -> xml_path_text_to_attribute xml_g_xpath_inside -> xml_path_node_inside xml_g_xpath_node -> xml_path_node xml_g_strdup_xpath -> xml_path_strdup // xml_path_flags_t 2002-11-11 "xmlg 0.5.3" <guidod@gmx.de> * xml/xmlgsave.* rename to xml_g_node_move_attribute_to_text * xml/pcresave.* derive an xml_g_node_save_attribute_to_text function that does leave the control characters in the text array and make these places chars in the attribute. This looks a lot nicer in an xml dump and makes it slightly better suited for pcre matching. However, this save routine is a lot slower than just doing memcpy in xmlgsave.* * make-check-attribute-saved added * xml/pcrecdoc.* adding xml_pcre_c_blocks to markup <block>s. also an example to use pcresave. 2002-11-11 "xmlg 0.5.2" <guidod@gmx.de> * xml/gerror.* fixing overwritten message * xml-test-error to test error handling * Makefile.am adding `make -C` generation rule for libxmlglib.la * xml/gerror.* use g_critical: g_error is always fatal *sigh* * xml/xmlgsave.* save a text area to some attribute - and of course its restore function to paste it back. That's best if some string literals would interfere with a pcre match on a text area. * xml/pcresave.* their pcre-path variants used for the two new test functions ./xml-attribute-saved-and-bank and its cousin ./xml-attribute-saved-restore 2002-11-10 "xmlg 0.5.1" <guidod@gmx.de> * xml/gerror.* and adjust all files to use xml_g_error calls. * xml/gstrfuncs.* to hold helper functions around charstrings. * ChangeLog added - needs to be maintained now 2002-11-09 "xmlg 0.5.0" <guidod@gmx.de> * moved lib sources to subdirectory "xml" * install lib headers to subdirectory "xml" * xml/gnode.* as renamed from xmlgnode.* * xml/gmarkup.* as renamed from pcremark.* * xml/gconfig.h instead of pkg-config.h - and install it along. * adjust everything to the new subdir "xml" * create rpm files - fixing relink error from libtool 2002-11-09 "xmlg prior" <guidod@gmx.de> 253333 Nov 9 21:17 xmlg-0.5.0.tar.gz 248334 Nov 9 03:00 xmlg-0.4.9.tar.gz 243637 Nov 8 15:28 xmlg-0.4.8.tar.gz 242032 Nov 8 14:16 xmlg-0.4.7.tar.gz 241065 Nov 8 03:27 xmlg-0.4.6.tar.gz 239921 Nov 8 00:11 xmlg-0.4.5.tar.gz 226668 Nov 7 06:08 xmlg-0.4.4.tar.gz 225742 Nov 7 02:00 xmlg-0.4.3.tar.gz 221376 Nov 6 20:20 xmlg-0.4.2.tar.gz 219541 Nov 6 20:20 xmlg-0.4.1.tar.gz 215318 Nov 6 05:21 xmlg-0.4.0.tar.gz 213173 Nov 6 01:13 xmlg-0.3.9.tar.gz 212311 Nov 5 04:16 xmlg-0.3.8.tar.gz 210538 Nov 5 00:35 xmlg-0.3.7.tar.gz 210337 Nov 4 22:02 xmlg-0.3.6.tar.gz 208557 Nov 3 18:46 xmlg-0.3.5.tar.gz 206990 Nov 3 05:10 xmlg-0.3.4.tar.gz 203711 Nov 3 01:17 xmlg-0.3.3.tar.gz 202563 Nov 1 20:44 xmlg-0.3.2.tar.gz 202501 Nov 1 18:01 xmlg-0.3.1.tar.gz 201531 Nov 1 06:13 xmlg-0.3.0.tar.gz 200830 Nov 1 04:58 xmlg-0.2.9.tar.gz 199799 Nov 1 03:08 xmlg-0.2.5.tar.gz 199160 Nov 1 02:06 xmlg-0.2.4.tar.gz 197821 Okt 31 22:01 xmlg-0.2.3.tar.gz 196847 Okt 31 18:06 xmlg-0.2.2.tar.gz 194934 Okt 31 04:12 xmlg-0.2.1.tar.gz 193136 Okt 30 22:56 xmlg-0.2.0.tar.gz 192639 Okt 30 21:40 xmlg-0.1.9.tar.gz 192536 Okt 30 20:18 xmlg-0.1.8.tar.gz 191336 Okt 30 17:21 xmlg-0.1.7.tar.gz 189904 Okt 30 03:25 xmlg-0.1.6.tar.gz 187762 Okt 29 23:17 xmlg-0.1.5.tar.gz 188522 Okt 29 18:18 xmlg-0.1.4.tar.gz 186896 Okt 29 10:51 xmlg-0.1.2.tar.gz 176765 Okt 21 03:02 xmlg-0.1.1.tar.gz |