Every feature

One workshop, 14 entries. Each says what it does, and what it does not do.

Open a file

Open an .mpp file

Drop the .mpp: its container is recognised, the format generation is named, and the way to get a readable schedule out of it is shown. The task detail of an .mpp is not read yet, and this page says so before you have lost five minutes.

The container is read in full: it is a compound file, the same structure as an old .doc or an old .xls.
The streams it contains are listed, with their sizes.
The format generation is named when it is recognisable. When it is not, the tool writes unknown rather than guessing.
A truncated file is refused with a message saying what is missing, instead of a half-finished silent read.
The way forward is shown: in Project, File, Save As, then XML format. That XML is read here in full.

What it does not do: The task, link and calendar detail of an .mpp is not read yet. That work is waiting for real files from several generations of Project, because a reader written from a single example breaks on the second one. In the meantime the Project XML is read in full, and it comes out of Project in three clicks.

Open a Project XML

Drop the file: the tasks, links, constraints, deadlines, calendars and resources are read, then the dates are recalculated. What the file stored and what the calculation gives are shown side by side.

Tasks, with their hierarchy, durations, progress, milestones and summary tasks.
The four link types (finish to start, start to start, finish to finish, start to finish) and their lags, positive and negative.
Project's eight constraints, its deadlines, and manually scheduled tasks.
Calendars: work weeks, dated exceptions, public holidays, and the calendars belonging to resources.
Resources and their assignments.
The dates stored in the file, kept as they are, and the recalculated dates, shown next to them.

What it does not do: The recalculation is not presented as identical to Project's. It is shown beside the file's dates, with the gap task by task and, where there is one, its explanation. Resource levelling is not performed.

Open a Primavera XER

Drop the XER: the activities, the work breakdown structure, the relationships, the calendars and the resources are read, and the schedule is recalculated. Primavera P6 is not required.

Activities, with their original, remaining and completed durations, and their type.
The project's work breakdown structure: each node becomes a summary task, and the hierarchy is rebuilt.
Relationships between activities, with their type and lag.
P6 calendars, including standard weeks and dated exceptions, read from the string P6 writes into the file.
Resources and their assignments.

What it does not do: Activity codes, user-defined fields and resource curves are not used. Klargantt does not write XER files back: the return trip goes through Project XML, a workbook or a CSV.

Open a GanttProject .gan

Drop the .gan: the nested tasks, the dependencies, the lags and the public holidays declared in the file are read, and the schedule is recalculated.

Nested tasks, with their duration and progress.
Dependencies, which GanttProject writes on the predecessor's side, put back the usual way round.
Lags, link types, and date constraints.
Public holidays declared in the file.
Resources and their assignments.

What it does not do: GanttProject's views, colours and custom columns are not carried over. Klargantt does not write .gan files back.

Convert and render

Convert a schedule to Excel

Drop the schedule, then take the workbook. Dates are dates, durations are numbers, and float is calculated. Sorting and filtering work, because nothing is written as text.

One row per task, with its identifier, level, name, duration and progress.
The dates stored in the file and the recalculated dates, in separate columns: they are never mixed.
Total float and free float, in working days, negative values included.
The critical path, hard constraints and deadlines, in columns.
The first row is frozen and filtered. A 5,000-row workbook is written in under a second.

What it does not do: This is not a Gantt chart inside Excel: it is the schedule data, in columns. For a bar chart, take the PDF or the image. The workbook is not read back by Klargantt: spreadsheet reading comes later.

Convert a schedule to PDF

Drop the schedule and choose the paper. The Gantt chart is cut into pages, the name column is redrawn on each one, and every page carries a dated footer and its number.

A4 to A0, portrait or landscape, or a single page at the exact size of the chart.
The name column is redrawn on every page: a page on its own stays readable.
A continuation marker says that a bar carries on onto the next page.
The footer carries the schedule name, the print date, the scale and the source of the dates.
The PDF is written here, byte by byte, with no third-party library.

What it does not do: The PDF is a print, not a working file: it does not import back into Project. The fonts used are the standard PDF fonts; a task name in a non-Latin script is not rendered yet.

Export a Gantt chart as an image

Drop the schedule, take the image. The PNG is drawn at twice the size so it stays readable once printed; the SVG stays vector and can be reworked in any layout tool.

The PNG is drawn on a canvas, at twice the size: at scale 1, task names become unreadable once printed.
The SVG carries the same shapes, in vector form, with selectable text.
Both carry the schedule title and a dated footer.
The time scale can be set to day, week or month.

What it does not do: An image is not a working file. For a paginated print, take the PDF; for sortable data, the workbook. The PNG is produced by the browser: it does not exist on the command line.

Convert a schedule to CSV

Drop the schedule, take the CSV. One row per task, the file's dates and the recalculated dates in separate columns, the float and the critical path.

One row per task, in schedule order, with the hierarchy level.
The stored dates and the recalculated dates, never mixed.
Total float, free float, the critical path flag, constraints and deadlines.
Predecessors, with their link type and lag.

What it does not do: A CSV carries no date format, no visible hierarchy and no chart. For a spreadsheet ready to sort, take the .xlsx workbook.

Recalculate

Calculate the critical path

Drop the schedule. Early and late dates are computed, float follows, and the critical path is drawn in red on the chart. The task holding the project finish is named.

Forward and backward passes, across the four link types and their lags.
Project's eight constraints, deadlines, manually scheduled tasks and progress are all taken into account.
Total float and free float, in working hours then converted to working days of the relevant calendar.
Negative float is shown as such, never clamped to zero.
Summary tasks carry the roll-up of their descendants, without being scheduled themselves.

What it does not do: The calculation is not presented as identical to Project's or P6's. It is shown beside the file's dates, with the gap task by task. Resource levelling is not performed, and an overload moves nothing.

Total float and free float

Drop the schedule: every task gets its total float and its free float, in working days of its own calendar. Negative float is shown as it is.

Total float: what a task can take without moving the project finish.
Free float: what it can take without moving its first successor.
Both in working days, converted from the working hours of the relevant calendar, not in calendar days.
Negative float, which says a constraint or a deadline cannot be met, is kept negative.
The quality check reports float above 44 days and negative float, as the DCMA assessment asks.

What it does not do: Float is read in context. Klargantt gives the number and says where it comes from; it does not conclude that the schedule is good, or that it is bad.

Holidays and work calendars

Drop the schedule: the calendars it contains are read in full, with their work weeks, dated exceptions and public holidays. The recalculation uses them, task by task.

The standard week of each calendar, with its working hours day by day.
Dated exceptions: non-working days, exceptional working days, special hours.
Resource calendars, which take precedence over the task's calendar when the resource has one.
Inheritance between calendars: a derived calendar takes what its parent says, and changes only what it declares.
Everything is computed in local hours and minutes, never in millisecond differences: a clock change moves nothing.

What it does not do: Klargantt applies the calendars found in the file. It does not add any, and it does not impose French public holidays on a schedule that does not declare them. A missing calendar is reported by the quality check, not silently replaced.

Check and compare

Schedule quality check

Drop the schedule: it is run through the 14 points of the DCMA assessment, plus three checks belonging to Klargantt. Each point gives its number, its threshold, and the tasks concerned.

Missing links, finish-to-finish and start-to-finish links, positive and negative lags, hard constraints.
Float above 44 days, negative float, durations above 44 working days.
Unresourced tasks, late tasks, the baseline execution index, the critical path length test.
Three further checks: declared calendars, the presence of start and finish milestones, and the legibility of the chart.
Four verdicts per point: met, exceeded, to be stated, not applicable. The report never concludes "compliant".

What it does not do: A report is not an opinion. Klargantt gives the number, the threshold and the tasks; it does not say whether the schedule is acceptable, and it does not replace the judgement of whoever signs. The thresholds are those published in the DCMA assessment; your contract may impose others.

Compare two versions

Drop two versions of the same schedule: the tasks are matched, then the differences are listed. Additions, removals, renames, date shifts, duration and link changes.

Tasks are matched in three passes: by identifier, then by path in the work breakdown structure, then by likeness.
Likeness combines the name, the duration and the position: "Foundations" and "Foundations (redone)" find each other, without confusing two tasks with the same name.
Each difference states its nature, its size, and how the match was made.
Added and removed tasks are listed separately from changed ones.

What it does not do: The comparison gives differences, never the cause and never the responsibility. It does not say whether a delay is anyone's fault: delay analysis is expert work, not a tool output.

Three-week look-ahead

Drop the master schedule: the three-week extract is taken around the status date. What is in progress, what starts, what has to be finished.

The window is set on the schedule's status date when it carries one, otherwise on today's date.
Tasks in progress, tasks starting, tasks due to finish inside the window.
The milestones of the period, which are what the meeting looks at first.
The extract comes out as text, as a workbook or as a CSV, and the matching chart as a PDF or an image.

What it does not do: The extract comes from the file you dropped. It does not replace site reporting: what the schedule says and what happened on site are two different things.